Today it’s just links and fun finds that I’ve been storing up for a about a week.

What the AIs thinks of the debate

I asked Perplexity about the US presidential debate last night. I wasn’t very interested in “fact checking,” but in how different the coverage in the New York Times and Fox News was. I added in a “partial transcript” of the debate as well to compare both to what actually happened.

The two outlets obviously watched a different TV show than the other, possibly broadcast from two different parallel universes.

The other interesting part was asking the AI what was not covered in each article: climate change is the biggest one, it seems. At the end, I asked it to make D&D characters for each. Overall, it was pretty good - at least enjoyable to read.

Anyhow, I also asked it to write an overview at the end. It was predictably boring in that it was devoid of any point of view. Meanwhile, here’s the view from across the pond in The Economist - plenty of PoV and style as usual.

Oh, and Gemini just flat-out chickened out and wouldn’t do anything. I tried to outfox it by saying it was a fictional transcript, but it was too wise.

(And, yes, obviously generative AIs don’t actually think anything about the debate, they’re just doing their thing outputting what seems like the word that follows the one before it.)

Relative to your interests

• Charles Schwab Adopts PostgreSQL (With VMware Tanzu) - Why Charles Schwab chose Postgres over Oracle, and chose to pay for Tanzu Data Services support.

• Cycle Time - Time spent figuring out this nuance is time spent not coding and getting your apps out the door and kicking in the product management feedback cycle. Still, good discussion of the nuances of the phrase.

• 2025 IT Budgets See 5.5% Increase as CIOs Invest in AI and Cloud - “With a projected 5.5% increase in IT budgets, the year ahead is set to focus heavily on artificial intelligence (AI) and cloud computing.”

• Executive translation - “Many high-agency managers try to prevent executives from doing silly things, but it’s almost always more effective to translate their energy for a silly thing into energy for a useful thing. It also leaves the executive feeling supported by your work rather than viewing you as an obstacle to their progress.”

• How Heroku Is Positioned To Help Ops Engineers in the GenAI Era - “We decided that the best thing for us strategically was to use Kubernetes underneath the covers, essentially to replace a lot of the code that we have been using to do the same thing. That’s a big migration effort for us. There’s a lot of expertise that needs to be built.”

• Oracle Runs OCI Clones At Rival AWS, Google, And Azure Clouds - Oracle runs its stack in the various clouds.

• Admins wonder if the cloud was such a good idea after all - ”According to a report published by UK cloud outfit Civo, more than a third of organizations surveyed reckoned that their move to the cloud had failed to live up to promises of cost-effectiveness. Over half reported a rise in their cloud bill…. Although the survey, unsurprisingly, paints Civo in a flattering light”

• Five Lessons Learned From a Lifetime of Platform-as-a-Product - Text version of a good talk.

• DOJ, Nvidia, and why we restrict monopolies - “A company getting it right does not give it some kind of permanent license to coast while printing money. The question that all antitrust seeks to answer is simple: how much is enough? When do the well-deserved riches and power that accumulate to companies that make big bets, execute well, and invest wisely start to be toxic for society or humanity as a whole, and for competition itself? Are the riches that the largest companies make earned, or are they simply the product of being large? If it’s the former, great! But if it’s the latter…”

• Platform Engineering Reshapes Software Dev at Bechtle - Customer case or Bechtle using Humanitec.

VMware Corner

What with our big conference, there’s been lots of coverage recently of VMware. Here’s some. I think the new strategy - focusing on private cloud - came across clearly.

• Under New Management: Impressions from VMware Explore 2024 - “Tan made comments about not chasing ‘bright shiny objects.’ The context around these comments – particularly from the analyst session – indicate Tan’s statements could reasonably be interpreted as pointed commentary about Kubernetes. The Tanzu platform has two available runtimes: Cloud Foundry and Kubernetes. More airtime was given to Cloud Foundry than Kubernetes, and the leadership team made statements about ‘focusing on existing customer success.’ All together, there were signals that read as the company focusing on nurturing the assets from the Pivotal acquisition over the Heptio acquisition, and prioritizing solutions centered around its VM-centric past more broadly.”

• Tanzu 10 Sets New Standard for Private Cloud-Native Platforms - ”According to a recent survey by Futurum Research, organizations that have adopted cloud-native technologies have seen an average of 20% faster time to market and a 15% reduction in IT costs.”

• Broadcom CEO: VMware still ringing the cash registers - 'That translates into 32% quarter-on-quarter VMware bookings growth." And: ‘When it comes to revenue tailwinds, AI is driving “about two-thirds in compute and one-third in networking” said Broadcom CFO Spears.’

• Broadcom VMware acquisition sees profitability amid disruption - “As long as [Tan] keeps the [total cost of ownership] for running a workload under the cost of the public cloud and under the cost of comparable competitors, he’s good,” Ellis said. “But the bar was low … essentially [VMware] was the discounted enterprise virtualization suite for Dell and EMC, and they used it to sell hardware; they used it to sell the other parts of their infrastructure. The price was artificially kept low, and essentially Hock Tan is allowing [time] to figure out, ‘Will it sustain this higher price?’ I think when you talk about the inertia, when you talk about how well embedded it is within the tech stack, the answer’s probably going to be yes.”

Wastebook

• “At the average organization, 56% of teams utilize a DevOps methodology. Organizations expect this to grow in 12 months by about 6 percentage points. However, it has been 15 years since the inception of DevOps, and organizations are still struggling to expand the adoption of DevOps across their application estates.” IDC: “Development Tools and Applications, 2024.”

• In tech, don’t confuse the story that drives your valuation versus the features that get customers to buy your products.

• Life is short. Don’t waste it eating chicken breasts.

• “tentacles of spend”

• “no Facebook employee has, to my knowledge, been killed by cannon fire” Here.

• And: “who remembers DR-DOS?”

• “We become Old, before we have been able to taste the Pleasure of being Young” Here.

• “Did any of your grandmother’s go to college?” RotL #551.

• It’s one dimensional chess at its finest.

• “galaxy-brain peanut gallerians” here.

Conferences

Talks I’m giving, places I’ll be, and other plans.

SREday London 2024, speaking, September 19th to 20th. SREday Amsterdam, Nov 21st, 2024. Coté speaking. Cloud Foundry Day EU, Karlsruhe, Oct 9th. VMware Explore Barcelona, speaking, Nov 4th to 7th.

Discounts! SREDay London and Amsterdam: 20% off with the code SRE20DAY. Cloud Foundry Day 20% off with the code CFEU24VMW20.

Logoff

In AI-land, I’ve been using Gemini and Perplexity more than ChatGPT. I re-signed up for Gemini and…it’s fine? I use Perplexity to the most for basic search and summarizing. Gemini isn’t that great at D&D - ChatGPT feels better. But, I’m still at that ceiling of usefulness for chat AI: without spending a lot of time, I can’t get the results I’m looking for. And that’s the point: it should be quick and effortless.

What I’m liking about Gemini is: it’s integration into the rest of Google (Gmail, docs, Google Drive, etc.) and that you get 2 TB of storage. This means I can use Google Photos again. ChatGPT can import Google Docs, and you can of course upload files, but it’s a lot different to just have Gemini right there in a document you’re working on, or in email. One theory: chat AI is just a feature, so without integrating with and/or owning the rest of the apps, it’ll quickly become less useful.

I used Gemini to help me figure out what to write about for a recent blog post, but it wasn’t good at doing the actual writing. Still, as an assistant it worked, this once at least. The “copilot” idea is a good one.

Gemini still can’t tell me the top five people I emailed in the 2000’s. It just says it can’t do that, despite having all of my email. Something like perplexity would figure out how to search all my emails, sort them by frequency, and then output that. I think?

I still think that the only people who think AI is going to change everything, is super-cool, are people who don’t use it much. The initial impression is amazement, but then you realize it’s just a dancing bear.

The DevOps community is running on fumes and at the lowest point in mindshare and interest that it’s ever been at. This is stupid. The practices, tools, and mindset of DevOps are vital to how most organization run their software1 and DevOps has improved the way the software we use everyday is built and run, improving all of our lives. If DevOps wasn't a thing, the world of software would be worse and each day would be a little more tedious because the apps we depend on would be worse. 

Here's what I think DevOps needs to do to come back from the dead:

(1) Steal back the flame from platform engineering

Developing software and building digital capabilities is becoming ever more complex. By 2026, 80% of large software engineering organizations will establish platform engineering teams as internal providers of reusable services, components and tools for application delivery - up from 45% in 2022. Gartner, circa 2024.

"Platform Engineering," is fine, but creating it as separate, new method of work has done a disservice to DevOps. Shoving DevOps into a grave was also a bit rude.

Platform engineering was a very well done marketing strategy to establish a new market for a handful of portal vendors to start working in. All of us joined the momentum of platform engineering. I have done it many times, and probably still will!

Much like The 12 Factor App, a piece of brilliant technical marketing, platform engineering has actually become its own, legit thing now. But, that early strategy of stealing the sun-light from DevOps made platform engineering seem like more than it is. Platform engineering is just DevOps with one small thing: adding in product management.

The rest of platform engineering is just adding new tools to the DevOps toolchain: primarily internal developer portals and all of the work and tooling that goes into making Kubernetes usable for developers. Over the next 12 months, platform engineering is going to take hold in "enterprises" (see Gartner prediction above), which is what I call large organizations that are not tech companies: you know, IT normies. There's early use here-and-there in enterprises,2 but not at the scale of DevOps, or even cloud native (you know, architecting your applications in containers).

The DevOps community has to take over, gobble up, or otherwise “ride” the platform engineering wave or get drowned by it, lost in the deep like some ancient city. It's too late to reverse the name. And, really, "platform" as a name is incredibly valuable and descriptive, and "platform engineer" is a fine synonym for "DevOps engineer."

The job title might even do a good job to clean house on what exactly a "DevOps engineer" is. This is the case with DevOps in general: it's become so dominate that “DevOps” has started to mean most anything, and everything related to building and running in-house applications. This is not good!

(2) DevOpsDays should be renamed to Platform Days 

When suggesting such a thing, one must always begin with praise.

The community-driven, global DevOpsDays organization is one of the most unique assets in the IT world. At the moment, there's nothing like it: a vendor-independent, volunteer-led organization focused on industry best practices that holds regional events to educate and enrich the community. 

Let me do some boosterism here:

1. Being vendor-independent is important for two reasons. First, it means the content is actually practical and "true" without the bias of a commercial agenda. Second, it means that all vendors feel comfortable sponsoring DevOpsDays. Vendors are incredibly important to the DevOps community. They're the ones who have funded its existence! If the overall community and DevOpsDays was owned by a single vendor or an oligopoly it wouldn't have been successful nor valuable. Vendors have an agenda to sell their products and their market category (people who sell on-premises software want to show buyers why private cloud is the best, people who sell public cloud software want to show buyers why on-premises software is better, people who have new tools of monitoring and log management want to explain to you why the old ways suck - shocker!). Vendor's budgets also come and go at the whim of their business, new technologies, and macro-economic headwinds. One year, the company is focused on growth at all costs to drive company valuation, the next year the company is focused on cutting costs because a PE firm bought them. One year, money is cheap, the next year money is scarce. One year, DevOps is important, the next year generative AI is important. Budgets shift! In short, depending on vendors to support an ongoing, multi-decade community is risky.

2. The volunteers are the ones, then, who make it all happen. The passion that DevOpsDays volunteers have for the DevOps community and event is extremely rare in other parts of IT. Even the agile community doesn't seem to muster enough energy to put on so many conferences. I don’t think application developers can be bothered to look up from their IDEs. There are regional meetups and user groups in IT, sure, but I don’t think there’s anything close to the scale of DevOpsDays. While my overly utilitarian, cold heart might ache at all the "culture" talk in the DevOps community, that vibe is incredibly real and important: it becomes part of the identity and lives of these volunteers. I don't want to see that lost, and it's a huge asset that the IT community should cherish and protect.

3. Focusing on best practices means that the talks, workshops, and discussions at DevOpsDays are focused on continually improving the IT community. There have been four, maybe even five stages of DevOps over the past 15+ years. The DevOpsDays talks focus on what the DevOps community needs help with at the time. You can see this in the community's shifting from automation (infrastructure as code, immutable infrastructure, Puppet, Chef, etc.), to monitoring and management (observability, chaos engineering, etc.), to culture (how people work, the mindsets that lead to good outcomes and good lives), to filling the usefulness-gap created by people's over-enthusiasm for Kubernetes, to the present.

4. DevOpsDays are in-person which is the most effective way to educate the community and keep it thriving. More than many other in-person events, the edge that DevOpsDays has is that the events are regional. They’re all over the world. This means that people in the DevOps community don't have to pay to travel, and if they do, it's usually just an hour or two to go to a nearby city. Compare this to one, large event a year in Las Vegas or Barcelona. The audience and community reach for DevOpsDays is so much larger and, thus, more effective.

The survival of DevOpsDays is more important than the survival of the name DevOpsDays. Because the platform engineering community successfully stole so much of the mindshare, they've sucked much of the sponsorship money out of the DevOpsDays. Worse than redirecting the money is drying the money up all together. Marketing budgets are tight now-a-days (I'm not sure why: vendors want to sell more than ever) so the people controlling event sponsorship money is looking for an excuse to cut costs. And, of course, there’s the redirection that platform engineering, drawing attention and therefore people away from DevOps.

There is currently no platform engineering set of conferences like DevOpsDays. PlatformCon is a good, online conference. I enjoy the recordings, have presented at them, and have encouraged other people to present. The PlatformCon crew is building up momentum for more in-person events (see all the regional group on their above page) - this is great! Good for them! Yay! RIP-TAYLOR-CONFETTI-BLAST!

But it will take years to replicate the reach and assets that the DevOpsDays community has.

I know it feels weird to the DevOpsDays people - it feels weird to me! - but renaming DevOpsDays to PlatformDays would solve a lot of problems quickly. Maybe the PlatformCon community could even merge with DevOpsDays, joining up the benefits of both. I don’t know. Maybe.3

(3) DORA should focus less on metrics and more on practices

DORA is probably the most important marketing asset DevOps has and has had. And it's a genuinely useful tool! DORA’s work is what helped spread DevOps into the mainstream, all those "enterprises.” And the annual cadence and DORA community has helped DevOps evolve, and kept it alive. 

The problem DORA has now is that most people focus only on the four DORA metrics. Common lore is that those enterprise normies don't look beyond the four metrics. This extends to DORA's annual clustering of elite, high, medium, and low performers: these are rankings. Those are just small parts of the overall knowledge in and utility of DORA. I don't think this short-sightedness of DORA users is intentional, or even conscious. It's how just people work. 

Whenever you put metrics and rankings in front of a corporate-human, eventually they'll focus most of their intention on upping their ranking. People do this because they think that being successful means being ranked high, getting a good score/grade. This is because that's how rewards are given in most organizations! Gaming metrics is the most rational thing for a person to do in most organizations. Each year, DORA refines high-grade fuel for this way of working and thinking.

Just as it's not really the fault of the users, it's not really DORA's fault either. But, DORA can help people do better by re-orienting their annual work.

First, the value of DevOps practices are not really in dispute. We all know that they're good. As with flossing, we all know that we should do the practices. DORA shouldn't stop gathering and analyzing these metrics, nor reporting on them. 

What DORA needs to do is move the focus reporting on the actual practices people follow. The activities they do day-to-day, the tools they use, that lead to the great results and clustering.

Of course, DORA currently cover this, but over the years the reports have focused more and more on a discussion of the survey and mechanics of the research rather than actions to take. This is fine! I get it - I have an engineering mindset and I would rather tell you about the internals of what I build, how I did it, and, most of all, what's cool about it. But the DORA-users need a tool. They need to know what to do, not why they should be doing it.

Sure, there's a danger that the users apply tools the wrong way, and you end up with a similar trap of focusing just on the four metrics. But, I mean: let's have a little more faith in people, and DORA’s ability to avoid that.

Let's also take a lesson from Extreme Programming. Extreme Programming is incredibly prescriptive. And, while I don't want to speak on behalf of the priests of that community, over the past few decades there's been a latent operating principle of using Extreme Programming: start with following all of the strict practices, learn what works and doesn't work in your organization, adapt the practices, and start the loop over again. That is how the DORA practices should be applied, and it should be the mission of DORA to make it happen.

In the most recent DORA report (2023) an easy to use discussion of these practices doesn't start until page 81 (out of a total report size of 95 pages). I confess my laziness here and apologize if it resulted in error, but after a few quick flips, I couldn't even find where the practices are covered after flipping through the 77 pages of the 2022 report, the 45 pages of the 2021 report, and had to go all the way back to 2019 to find this on page 31 of 82 pages:

…and page 57 to find the equally valuable this:

There's a snark in the DevOps community that the problem with Accelerate is that everyone who read it came across the four metrics on page 17, declared that they had all they needed, and then closed the book. Well, how about we put a discussion of practice - these very charts - on page one, or, at least page 9.

People in enterprises who are struggling to get better at software need this kind of material: easy to understand, focused and quick, and possible to put into place practices. How do you ensure you floss everyday? First, buy some floss. Second, put it where ever you brush your teeth. Third, notice when you do it and don’t do it. Fourth, keep trying to do it better each day. Life-hack: put it on your desk and turn off the camera when you're bored in a dumb Zoom meeting you can't get out of, and floss. You could also keep the camera on.

The "model" (as I think DORA calls it) is there, and it's even nicely done on the DORA community site (see the figure at the beginning opening of this section). What I'm suggesting is that it should the primary, the first focus of each report. When you email around a 45 to 95 page PDF in enterprises you have one shot to get people to read it and benefit from it. They're looking for the most effective tool they can use right now to achieve whatever goals they have. That practices chart is the tool they're looking for and it should be the first thing they see in the 30 seconds they're going to spend on that PDF. The rest of the work is great, valuable, and hard to do, but those are the "backup slides" that you'll use once you get people's attention and show them what you're proposing.4

We don't need further proof that flossing is good, nor to be told that people who floss have elite teeth. We need to know the mechanics of flossing and how to build a system that lets us put in place continuous flossing.

(4) DevOps should add product management for infrastructure

I give platform engineering a hard time because the people who launched it tried to push the DevOps community in the grave. Or, I don't know, maybe they were just having fun with the headline. Har, har, har…good one.

But, rasing the prominence of product management is an incredibly valuable thing that the platform engineering community has done. And it’s why I think it’s a good, helpful body of work.

The ideas of product management - identifying your customers/users and using, their problems/jobs to be done, crossed with what is technically feasible, and using feedback loops to continuously build a product that best solves their problems…with software - haven't been a large part of the DevOps community. The platform engineering community frames their work as building platforms for application developers, treating them as "customers," and, therefore product managing infrastructure for application developers.5

Sure. Absolutely. Without a doubt. The DevOps community has wanted that outcome and worked towards it. It's right there in the name: Dev! But product management as a practice hasn't been a focus in the DevOps community. It should be! There is a-funny-because-it's-true joke that there are very few devs at a DevOpsDays conference.6 The same is true for the community as a whole.

The good news here is that product management is one of the most understood, well documented, and proven methods of working in software. Thus, from a regular person's perspective - those people enterprises - product management is straightforward and learnable.

Despite product management being easy enough to learn, I think it’ll be incredibly difficult for enterprises to add product management to their infrastructure/DevOps/platform groups. They’re going to need help, and the DevOps community has a proven track record for such tough work.

Some of the people who introduced product management into infrastructure ("platforms") over the past ten years have done a great job grafting their term of art, “platform as a product,” into platform engineering. And it's stuck! I wish this had happened in the DevOps community, but because the platform engineering fig vines took over, it didn't.

Well. You know. We should bring product management into the DevOps world.

For example, DORA could start incorporating it into their studies. While product management is understood for applications, there's not a lot of work on applying it to infrastructure (yes, again: "platforms," if you prefer) and developer tools. Us vendors know about it since it’s our business, but figuring out how to do it in enterprise infrastructure (or even apps!) isn't much studied. DORA can do that, and it'd help clear out some of those vines if they did. (I know the Puppet study-surveys are going for it, so good on them!)

In fact, I'd go as far to say that product management is the missing piece for the wider success of DevOps.

(Now, if you wanted to be clever, you would think this: platform engineering is focused on developers and product manages for them. DevOps will do that, of course, but what a DevOps person will tell you is operations people are also customers [and maybe security people if they behave]. Why not product manage for ops too?)

I will be sad

Listen: I know! I hear you.

But, I'm not saying your efforts are wasted - it's all good stuff. We just need to do more. We need to adapt and change. Fast feedback loops, continuous learning, and changing to do what works are core principles of DevOps. This is what I'm seeing in the feedback loops and the theories I have for how to get better.

I give a huge, massive, rainbow-smelling shit about the DevOps community. It's been a positive force in my career and life. Many of my friends are in the DevOps community and going to DevOpsDays is how I see them. It’s fun! DevOps has also had an important, large impact on how the world builds and runs software. And not just “hats on cats” apps, but apps that you and use everyday to run our lives.

There's a lot of existential malaise going on in the DevOps community, especially among the most enthusiastic and committed members of the community. Some (many?) of the original DevOps community members have drifted out of the community, and will sort of roll their eyes over the whole thing over drinks. There isn't a better community in IT than the DevOps community - it's managed to remain kind, helpful, and nice. I'll be sad if we can't keep it alive and thriving.

Logoff

I guess what I’m trying to say is: DevOpsDays Antwerp was good. It was nice to visit with everyone and see some great talks.

Searching for Cheap Tricks

The amount of AI content and conversations out there is getting exhausting. It’s almost as bad as the burbling font of platform engineering content that gets rewritten and published each week. The below RAND paper and commentary got me thinking though: a lot of the AI talk is just talk about applying new technologies in general.

We are still really bad at the basic task of communicating requirements to developers, checking in on their work and seeing if they’re solving the right problems in a helpful way…and even knowing what business problems to need attention in the first place.

Here’s an email I wrote to a friend on this:

Re: my prodding to figure out cheap tricks for getting organizations to improve how they use computers, this paper has a few theories of why AI projects fail. I feel like you could do a search and replace for "AI projects" to just "IT project" and the same could be said. The core, non technical problem is two sided:

(1) getting non-technical leadership to understand what AI can do, how much effort it takes to do the work, and most importantly how to properly set the requirements: making sure you solve the right problem in the right way. This last part is the mystic "understand what you're doing, the point of it, the why's of your business." Like we were talking about last night in the backstreets of Antwerp, my theory here is that if you asked most people in a business "how does the business operate and function each day?" they would have poor answers, let alone answers that an AI could do anything with, or that programmers could do anything with.

(2) the technical people are unable to "speak the language of business" and get the people in #1 to "speak the language of IT." I'm sure this is a lack of communication, facilitation, and interpersonal skills ("managing upwards"), but it probably results in the tech people not pushing back enough and drilling down to get the non-tech people to be helpful. This is the pesky thing of consultants always saying "before we talk solutions and projects, let's take a step back and examine what you really want." People are eager to fix/improve the problems, and don't often question if they're solving the right problem, nor understand the tools available to solve it. (Including the tool of continuous learning which would say "we have to use a tight-feedback loop to learn how to solve the problem).

It all amounts to introducing an ongoing "culture" of finding and understanding the problems you're solving, and a similar process of "leadership" frequently checking in to see if the tech people are getting closer to solving a useful problem, or if they're straying away from the original requirements.

One of original goal of agile software development was to address these problems: “business IT alignment” as we sometimes called in the 2000s.” On the one end, Extreme Programming had a concise, clear cookbook to follow. I like these prescriptive, “do this” approach to corporate improvement. They’re a collection of “cheap tricks” you can start applying.

However, the “fix” is to change the mindset of the company, from management to worker, to always be curious and on the hunt for how to do things better, “continuous improvement.” There isn’t always a way to do things better, and you also have to spend most of your time actually doing things.

But, there’s at least two things that most organizations could be better at:

1. Understanding what they do, the job to to be done, and how their organization does that - I have a feeling that if you asked most people, tops to bottom, at an organization “how do things really work around here?” you’d get mixed results, most of them only part of the overall picture. Can you draw out and end-to-end flow of all the activities it takes to deliver the product or service to the customer?

2. Spotting when to experiment with changes. This is applying Toyota kata and all that. The mystic version of “continuous learning” is that you’re always on the lookout for learning and improvement. But, you also have to actually do the work. You have to check out people at the grocery store, put the Fedex letter into a person’s hands, send out a plumber at the right time to the right place to unclog a sink (and then they need to unclog it), transfer money between two accounts, etc. When is it OK to stop “doing things” and work on improving things? That’s a very hard “stop the line” call to make.

I want to find a way to combine the mystic management mind-games of most corporate improvement and transformation talk with “cheap tricks” that get people to start improving and keep improving.1 This is something like “the habit of building habits.” It’s all well and good to talk about “habit stacking” and tell people that they need to establish new habits and routines. Yes, but how do I start the habit of starting habits?

Once the people in an organization have gone through the mystic experience of asking “why” or even discovering their customer’s milkshakes, what are the daily tasks and routines they do? What tools do they use? How do they actually do the work? What are the cheap tricks they can start using to be better?

Relative to your interests

• Moonbound Revisited - “Several decades ago, the semiotician A. J. Greimas claimed that all stories are comprised of six actants, in three pairs: Subject/Object; Sender/Receiver; Helper/Opponent.”

• The Forgotten Surrealist Painter Who ‘Didn’t Have Time to Be Anyone’s Muse’ - “In contrast to the tendency of Surrealism’s male artists to depict women as thin, young, fragmented, static, and perpetually naked muses, Carrington’s women fell across a spectrum: often very old, powerful and threatening, in a state of action and transformation.”

• How Platform Engineering Enables the 10,000-Dev Workforce - Vendor survey overview.

• How the buildings you occupy might be affecting your brain - “Could bad architecture also be contributing to the development of neurodegenerative and psychiatric disorders?”

• Is AI a Silver Bullet? - “This is the trade off we have seen before: eliminating the accidental complexity of authoring code has less value than we think because code spends most of its time in maintenance, so the maintainability of the code is more important than the speed to author it.”

• The Root Causes of Failure for Artificial Intelligence Projects and How They Can Succeed: Avoiding the Anti-Patterns of AI, RAND paper - As ever, communicating the requirements and the problem to be solved to IT is difficult and often results in solving the wrong problem, at least in the right way: “In failed projects, either the business leadership does not make them- selves available to discuss whether the choices made by the technical team align with their intent, or they do not realize that the metrics measuring the success of the AI model do not truly represent the metrics of success for its intended purpose. For example, business leaders may say that they need an ML algorithm that tells them the price to set for a product–but what they actually need is the price that gives them the greatest profit margin instead of the price that sells the most items. The data science team lacks this business context and therefore might make the wrong assumptions. These kinds of errors often become obvious only after the data science team delivers a completed AI model and attempts to integrate it into day-to-day business operations.”

• Why “AI” projects fail - “And we’ve been doing this for decades now, with every new technology we spend a lot of money to get a lot of bloody noses for way too little outcome. Because we keep not looking at actual, real problems in front of us – that the people affected by them probably can tell you at least a significant part of the solution to. No we want a magic tool to make the problem disappear. Which is a significantly different thing than solving it. Because actually solving a problem includes first fully understanding the reasons for the existence of the problem, reasons that often come back to an organizations’ structure, their – sometimes inconsistent – goals and their social structure in addition to actual technical issues.” And: “AI is so good at writing summaries of all the meetings I can then have summarized afterwards to understand what was going on” says that your organization has probably too many meetings that are not structured well and that you don’t think that understanding why people say a thing, how they present their case is important (which makes you a bad manager).”

Per Request

Wastebook

• “Progressive doomerism.” Here, but, interestingly it was my perplexity.ai summarizer that came up with that phrase which did not exist in the article.

• “I want to stay little like this. I don’t want to get bigger.” My four year old.

• “Dystopian hod-rod races.” Apple Music summary of Liquor in the Front.

• Similar to “resume-driven development”: “Given the demand for new AI projects, interviewees reported prioritizing projects that grab headlines, improve reputation, and increase institutional prestige. We refer to this as activity prestige, which is the amount of positive attention given to some projects based on the public demand for those projects' outcomes.” Here.

Logoff

I’m in Antwerp for DevOpsDays Antwerp. This is the 15th year anniversary of DevOpsDays, so I wanted to attend. I’m lucky that they accepted a five minute talk from me as well. In the talk, I’m trying to answer the question “has DevOps made things better?” I do that by looking at changes in the DORA clusters, but more importantly, changes in how those clusters are made (thanks to some help from Steve Fenton and Nathen Harvey understanding how that is all done and, thus, a more accurate y/y analysis).

And, then, of course, I give some advice on what the DevOps community should focus on now. These are just two points I talk about a lot here: (1) make sure you actually have build automation in place, because it seems like many people don’t, and, (2) stop building new platforms and just use pre-built ones. If there was a third common theme for me, it’d probably be something like “start the practice of product management” this despite how difficult, even impossible, that seems.

//

There are two Brueghels in Antwerp, at the Museum Mayer van den Bergh. First, it’s a great museum if you’re into Flemish and Dutch golden age paintings. The building itself is a nice Netherlands canal house style building. The two paintings are Mad Meg and Twelve Proverbs.2 Here is Mad Meg:

Seeing pictures like these is much different in real life than online or even in books. It’s not that much different so as to be divine or anything: you just see some new things like textures and your field of vision either gets bigger or smaller. For example, with Mad Meg, you can really get in there and study the countless micro-scenes. In contrast, with Twelve Proverbs, you can see how basic and simple the pictures are; unlike the Brueghels most everyone knows, these pictures are not chaotic and “alive.”

Anyhow.

If you like that kind of thing, you should check out on of my favorite books, Short Life in a Strange World. If that book is a good fit for you, while it’s a bit mystic, it’ll help you enjoy “art” in new ways. If you just want the cheap tricks, check out the equally good Art as Therapy.

Just links and fun finds.

Relevant to your interests

I didn’t do link round-ups this past week, so there’s a backlog. Also, I’ve been using Perplexity to summarize some articles (not listed here, usually). I think it’s pretty good, and I recommend you use it.

• Amazon’s CEO says their AI tool has saved them a crazy amount of time - This oddly specific, and a big deal if applicable to other organizations. // ‘The average time to upgrade an application to Java 17 plummeted from what’s typically 50 developer-days to just a few hours," he wrote. “We estimate this has saved us the equivalent of 4,500 developer-years of work (yes, that number is crazy but, real).” The AI is not only fast but seems pretty accurate, too, according to his post. Amazon developers shipped 79% of the AI-generated code reviews without any additional changes, Jassy wrote.’

• How Lidl accidentally took on the big guns of cloud computing - A sort of community cloud for Germany and Austria? // “Its IT unit, Schwarz Digits — which became a standalone operating division in 2023 — has signed up clients including Germany’s biggest software group SAP, the country’s most successful football club Bayern Munich and the port of Hamburg. Last year, the unit generated €1.9bn in annual sales and it employs 7,500 staff.”

• 3 ways CIOs can turn strategy into execution - “Less than half of enterprises reach the majority of the strategic goals IT leaders set out to achieve, according to Gartner research.”

• How Platform Engineering Enables the 10,000-Dev Workforce - “60% of companies in the survey said they are releasing code on only a monthly or quarterly basis”

• Does Market Share Still Matter? - "we find that the positive relationship between market share and profitability generally still holds, but that it weakens with greater digital transformation. This means that market share increases translate into lower profitability gains for highly digitalized compared to less-digitalized companies. Additionally, it implies that digital transformation benefits smaller firms in particular because it leads to greater relative profitability gains for them compared to larger firms by removing some of the limitations that traditionally hindered them from becoming more profitable.” // “This means that the more digitalized a company becomes, the less upper management should rely on and prioritize market share to grow profitability. And the smaller a company is, the more it should prioritize digital transformation to boost profitability.” // See also the original paper.

• Dell’s AI Server Business Now Bigger Than VMware Used To Be - Maybe private AI will really be a thing: ‘“We are still in the early innings, and our AI opportunity with tier 2 CSPs, enterprise, and emerging sovereign customers is immense,” Jeff Clarke, Dell’s chief operating officer, explained on the call with Wall Street analysts. “Our view is supported by an AI hardware and services TAM of $174 billion, up from $152 billion, growing at a 22 percent CAGR over the next few years. We are competing in all of the big AI deals and are winning significant deployments at scale.”’ And: more analysis of Dell’s numbers.

• analyze_interviewer_techniques - “You are a hyper-intelligent AI system with a 4,312 IQ. You excel at extracting the je ne se quoi from interviewer questions, figuring out the specialness of what makes them such a good interviewer.” And, an example of it in action with a Tyler Cowen interview. Great looking stuff!

• Broadcom has brought VMware down to earth and that’s welcome - ”I left VMware Explore feeling that Broadcom has brought VMware down to earth and that’s a good thing, and that some of the complaints about the current state of VMware are a very natural struggle to cope with sudden and significant change.” // Great example of what a column on covering a tech conference should look like.

• VMware Tanzu Platform 10: Unified cloud app development simplified - ‘"With Tanzu platform 10, we have a unified single console to manage your applications, look at your CVEs, security governance, compliance,” she stated. “There are three main messages. Keep it very simple for the developer, never expose YAML files, configuration files and lower level configs. Number two, make sure that the deployment is dynamic and it automatically adjusts to the app needs. Then the third one is you want to do continuous update, repair and continuous security.”’

• A Java Language Cumulative Feature Rollup - ‘I found myself asking myself the question, “What’s every new feature Java has introduced since the last time I really cared about new Java language features?”, and didn’t find an easy answer via Google. So, I decided to create that list.’

• Is Your Organizational Transformation Veering Off Course? - “The research highlights that shifts in emotional energy, such as increased frustration or anxiety, can signal that a transformation is off track. Addressing these emotional shifts can help prevent derailment.”

• Being quietly radicalized by being on holiday - Pretty accurate view of the European lifestyle. “Enjoying life” is part of the ROI analysis.

Wastebook

• “There’s a fox a chicken and a bag of grain and now we want an ice cream so you go first and take the fox then leave the chicken but swap them with a soft-serve in a cornet with a flake.”

• And, the title itself: “Beaches are for people who enjoy the bureaucracy of going to the beach.”

• A very elegant tourist trap.

• If a PowerPoint joke is worth doing, it’s worth doing right.

• “Tip Jar Culture” and the “Tip Jar Economy.”

• If you find yourself tsk’ing at something, that’s a good indication that it’s something you should not care about. A tsk is a second arrow.

• Interacting with AI is like slot machines: you win just enough times to make you pull the lever again.

Logoff

As mentioned above, I’ve been really liking Perplexity. I use it for summarizing articles, but it’s also really good at what we now call “search.” Either asking quick questions that I’d other wise google-hunt for, like how to do things in apps. You know “google-hunt”’ing: you search for something in Google and spend time sorting all the crap and RTFM jerk’s answers to find a simple answer. Not with Perplexity! Sure, it takes longer than that “instant” Google search, but you save time on the second part of a Google search: filtering out all the bullshit. Plus, you can narrow down and integrate Perplexity more.

The Collections functionality is a different approach on making GPTs. You setup a persistent prompt that’s used in each new chat in the collection. So, for example, I have a collection that’s called “tl;dr,” and I have a “summarize this for me…” style prompt. So, each time I start a new chat in that Collection, it uses that prompt.

I need to experiment with using it for non-search chats, like playing D&D. I’m not sure it’ll be good at that kind of inward looking, hangout with the AI thing like ChatGPT is. On the other hand, I bet it’d be a good therapy-bot because it can go and search for things to add context.

Theory: that’s something that makes Perplexity good: Context. It can use Context (somehow) to weed out all that RTFM jerk stuff (this is sort of anti-Context - the people writing that stuff are operating in a different context than you are), and because it can search the live web (right?) it can add in relevant context to whatever you’re discussing with it.

Anyhow, you should use it!

I signed up for the annual plan. Sure, ChatGPT says they have search coming out, but (1) I don’t believe them, they announce all sorts of things and then trickle them out over the next year, if at all, and, (2) Perplexity is great!

//

School starts next week, so Kim and I are on one last, just-us weekend trip thanks to my mom’s annual, summer visit.

Normally I wouldn’t disclaim this since I think you, dear readers, are wise enough to know that it always applies, but: these views are my own, not my employer, VMware Tanzu by Broadcom. Also, we covered the below on this week’s Software Defined Talk. If you prefer to listen a podcast, it’ll be out on Friday morning, 7:30am Amsterdam time: subscribe!

Private Cloud

Yesterday at my work’s big conference, Explore, there was a lot of conversation about private cloud. It’s pretty straightforward and clear: VMware is your enterprise private cloud platform. From Hock Tan, Broadcom CEO:

“Public Cloud is much more expensive, more so than you ever expected. Complexity – another platform means another extra layer for you to manage. And compliance; you have regulatory policy requirements. It’s more complex, it’s more expensive and compliance is hard,” Tan told attendees. He said many CIOs are now moving their workloads back on-premises to deal with these three Cs, which represents a huge change, but also a huge opportunity. “Here’s my view,” Hock said. “The future of the enterprise is private. Private cloud, private AI fueled by your own private data. It’s about staying on-premises and in control. Of course, you’ll continue using the public cloud for elastic demand and bursting workloads, but in this hybrid world, the private cloud is now the platform to drive your business and your innovation, and we have work to do to make that happen.”

In round-table commentary, analyst Dave Vellante adds in the “AWS of the private cloud” angle. Commenting on the analyst day discussion with Hock Tan, Dave says:

I mean, he said in the keynote today, you basically get AWS on-prem. In the analyst briefing he said, okay we’re gonna to take 30, the top 30 services in the in AWS and we’re going to replicate those. We’re not going to do 300. But we’re going to basically create a substantially equivalent on-prem experience to the public cloud. And that’s their entire strategy. And you know I think if to the extent that they do invest in that road map people are going to stay with it why would you rip and replace VMware all?

Figuring out how many workloads are in private cloud versus public cloud is one of my ongoing side-projects. It’s not as easy as you’d think! Estimates are all over the map. 50/50 is probably a good estimate based on analysts, surveys, and other estimates I’ve seen in recent years. If you like something lower, Daniel Newman of The Six Five/Futurum estimates that only about 20% to 30% of workloads ever migrated to public cloud in the first place.

It’s also hard to find numbers around how many workloads are moving back to on-premises from public cloud, “repatriation” they call it. VMware sponsored an IDC white paper that pulls from IDC cloud surveys, saying this:

Two-fifths of respondents to IDC’s recent U.S. Cloud Migration Survey indicated that they performed or initiated repatriation of workloads from public cloud to dedicated environments in 2023, and 42% said they plan to repatriate some workloads in 2024. Security, data privacy, cost management, and performance are the top reasons companies engage in repatriation activities.”

IDC ROI PDF sponsored by VMware, August, 2024.

Here’s a similar data point from Forrester in 2023:

According to Forrester’s Infrastructure Cloud Survey in 2023, 79% of roughly 1,300 enterprise cloud decision-makers surveyed said their firms are implementing internal private clouds.

Here’s a very DevOps-y idea discussed in another Explore interview: most enterprise IT departments are run very inefficiently, “culture” and organization wise. There are silos for everything, and they are “best of breed,” which actually means the downside of locally optimized: different groups use different tools, you have lots of wait between handoffs, turf wars - all of the things people don’t like about large bureaucracy. The suggestion being that if they focused on organizing better, and consolidation, they’d get the gains that public cloud gives them. After all, that bureaucracy optimization is a side-effect of using public cloud: you no longer have such a large, bespoke IT department. You just have one of the public clouds and how they do things, take it or leave it.

Tanzu Platform 10 - Cloud Foundry and Kubernetes In One Runtime

I work in the Tanzu division, so I’m interested in what’s going on there. Here’s the big wrap-up post of everything.

Here’s my take on Tanzu.

Our main thing is this: the Tanzu Platform is an enterprise-grade PaaS. The primary focus is on private PaaSes (see above!). This can mean running on-premises, or running it in the public cloud with control over your own, single-tenant PaaS in public cloud.

People don't really say "PaaS" anymore, which I think is weird. They say "platform," so I'll use that word here.

The Tanzu Platform1 is composed of two approaches to building a platform:

1. Tanzu Platform for Cloud Foundry2 - a platform with a completely pre-built, turn key PaaS. This is Cloud Foundry, used by many large enterprises, supporting thousands of applications. In this case, you don't build the platform, but you can customize it here and there and add in services (“middleware”) that you need. You trade the ability for infinite customization for stability, consistency, and saving time on building and maintaining your own platform. You get those bureaucracy-busting benefits of going to public cloud.

2. Tanzu Platform for Kubernetes - a framework for building a PaaS on-top of Kubernetes. Added to this are tools for managing the apps running on these runtimes. Also, developers and platform engineers have the consoles needed to manage those apps. In this case, we provide defaults and templates (“opinions” to use PaaS-parlance) but you focus on building the platform to match exactly what you want, and can do so. We layer more usable yaml on-top of Kubernetes yaml, and have a good GUI as well. You can get a sense of what this all is by poking around at our docs on the Tanzu Application Engine.

3. There are data services built into Tanzu Platform as well: Postgres, MySQL, Redis/Valkey, Rabbit, Greenplumb (data warehouse), GemFire (in-memory data grid

4. And, the Tanzu Platform has very tight integration with Spring. Most enterprise applications are written in Java, and many of those use Spring. This means that Tanzu Platform is a great fit for most enterprise applications.

5. And, you know, all the runtime, management, consoles, integrations with infrastructure, etc. stuff you’d expect for a cloud platform.

When you spell it out it seems like a lot, but that’s sort of it. There’s of course a lot of details with such a wide scope. But, you can really simplify it as: the Tanzu Platform is a PaaS that will run Cloud Foundry or Kubernetes. The rest is just details in the yaml.

Here’s more from an interview with the GM of Tanzu, Purnima Padmanabhan:

The first part of Tanzu 10 is: we are bringing the reinvestment and the power of Cloud Foundry back to the Tanzu Platform. And Tanzu 10 brings in a single umbrella the choice of application runtimes. So you can build your application once and deploy either to a Cloud Foundry based application infrastructure, or application platform, or a Kubernetes either one, all within a single platform.

(~0:43)

The return of the Cloud Foundry philosophy and principles:

[There are ] three main messages. [Number One] Keep it very simple for the developer. Never expose yaml files, configuration files, kubectl, and lower level configs, BOSH, none of that. Just give them simple interfaces, say build my container and deploy, that’s it.

Number two, make sure that the deployment is dynamic and it automatically adjusts to the app needs. You just say, I want HA. You don’t have to tell me which cluster, how to connect the network, how to tie it all together, we will take care of it.

And then the third one is, you want to do continuous update, repair and continue security.

All these are actually pages from the Cloud Foundry world that now we are bringing to Tanzu platform as a universal tenant that applies across everything, whether you deploy in Cloud Foundry or on Kubernetes. (~11:00)

There’s something important to notice/know about “Tanzu Kubernetes”:

The core Kubernetes runtime is now baked into VCF, which is called the Tanzu Kubernetes Grid, it is made available with VCF, just like public clouds have their runtimes: EKS, GKS, AKS. Tanzu Platform layers on-top of that TKG runtime and gives you the application platform, the ability to build your application fast, the ability to deploy it with a single command, the ability scale it, the ability to secure it.” (~1:50)

This is something that’s still not widely understood: while VMware’s Kubernetes “distort” is still called “Tanzu,” it’s actually part of VCF. The Tanzu Platform requires an existing Kubernetes “dial-tone.” That is, you have to bring the Kubernetes. Of course, per what Purnima says, if you have VCF, you have Kubernetes. This is a minor distinction in the big picture, and maybe, actually, not really important to customers. But it’s worth keeping track of if you, you know, are into these kinds of industry details.

More

• My round-up of the SpringOne content from Monday.

• Here’s a recording of the Hock Tan keynote.

• Tanzu and Spring blog posts: Spring Announcements Blog; Tanzu AI Announcement Blog; Cloud Foundry specific Announcements; Data Announcements Blog; K8s specific Announcement Blog; Tanzu Platform Roundup Summary of Announcements Blog; Tanzu Sessions at Explore.

Conferences

Talks I’m giving, places I’ll be, and other plans.

This year, SpringOne is free to attend and watch online. Check out Josh’s pitch for the event. There’s an on-site conference as well at Explore if you’re interested. But, for those who can’t, now you can watch all the fun!

SpringOne/VMware Explore US, August 26–29. DevOpsDays Antwerp, 15th anniversary, speaking, September 4th-5th. SREday London 2024, speaking, September 19th to 20th. SREday Amsterdam, Nov 21st, 2024. Coté speaking. Cloud Foundry Day EU, Karlsruhe, Oct 9th. VMware Explore Barcelona, speaking, Nov 4th to 7th.

Discounts! SREDay London and Amsterdam: 20% off with the code SRE20DAY. Cloud Foundry Day 20% off with the code CFEU24VMW20.

Logoff

That should be it for my Explore coverage. There’s a lot of better coverage out there. I’ll be back with links and fun finds next time.

SpringOne

Yesterday we put on Spring One. There’s still sessions going on, which you can watch live if you register for free. Here’s some highlights.

We went over one of the most valuable products the Tanzu team has put out this year: the Spring Application Advisor. You set this up in your pipeline and it continuously scans for Spring and Java components that need to be updated. Using OpenRewrite recipes, it then creates the code you’d need to apply to not only update those components, but also update your own code and configuration.

This is a huge amount of work to do manually - Amazon estimates that it takes about 50 days. Just generating those changes is huge, but then also having the workflow to look at them, approve them, and make them official is a big help too. After the first go at it, you can then set up your CI/CD pipeline to automatically do this checking.

In large enterprises, keeping your Java, Spring, and other libraries up to date is difficult. Well, it’s more like a vicious cycle where people don’t prioritize it and get stuck using a mash-mash of years (and years!) old frameworks and components. This means they get stuck in the legacy trap: they need to upgrade, but they have no idea if it will break things, so they just keep using the old versions.

In a similar vein, Tanzu Spring also includes the Enterprise Spring Boot Governance Starter which puts in enforces various security and policy standards, like FIPS and PCI. Check out this overview as well.

Spring AI

Of course, what you really want to know about is the AI stuff.

There was a great demo of just how easy it is to add generative AI to a Spring app. Josh Long, Cora Iberkleid, and Mark Pollack went over a demo of doing just that. They made it seem effortless, and that’s the point of using Spring AI. In addition to just wrapping APIs around common AI calls, there’s also common patterns, defaults for setting up chat and RAG…in summary, a ready to go Java stack for using AI in your apps.

There’s also, of course, deep integration between Spring AI and the AI services in the Tanzu Platform. This means that if you’re using the Tanzu Platform for Kubernetes (tPCF), you can start messing around with AI now. There’s all sorts of models that work with it if you want to do some “private cloud” AI, and you can also call out the OpenAI APIs and others as well.

I haven’t watched it yet, but my co-worker Nick did a talk demo’ing Spring AI + tPCF. He walks through a couple of demo apps. This shows the coding, of course, but also how it all integrates together in the Tanzu Platform. He goes over the a platform engineer would do to add AI services to their platform, here our Cloud Foundry PaaS. And, there’s more details on Spring AI here.

Tanzu Platform

Today’s focus at the conference is on our PaaS, the Tanzu Platform. I’ll write more about this later this week. In the meantime, here’s the big round-up, and blog posts on our Kubernetes-based PaaS, our Cloud Foundry based PaaS, and the bundle of databases, messaging systems, and other data stuff we have.

Logoff

In an effort to find any videos from Explore in YouTube, I came across this recording of the original launch of Cloud Foundry in April, 2011. It’s interesting to ponder how far - or how not far - the industry has gotten since then. I think you could change a few words around (put in “Kubernetes” where “Antiquated Middleware Stack” is in the above - not that Kubernetes is antiquated [it’s only 10!] but that it causes stack-complexity problems), update the clothes, and you’d have pretty much the same situation and general pattern of fixing it as you saw back then. We keep going back and forth with the platform build-out.

Two major items of note: I don’t think ruby ever took off enterprise-wise. You can see how much they try to legitimize ruby for enterprise use. Back then, I guess, open source was still sort of scary for enterprises, and definitely not the way a lot of infrastructure was done. There’s a lot of open source in the enterprise boosting that you just wouldn’t see now: it’d be taken as the default.

This week’s Software Defined Talk podcast

We talk about the open source “rug pull” du jour in this week’s Software Defined Podcast: “This week, we discuss CockroachDB's relicensing, the ongoing debate about remote work, and platform engineering. Plus, some thoughts on the use of speakerphones in public.” We both steer us towards a conclusion something like: yup, the open source vendor can change the license and start charging whenever they want, that’s the new reality. You can control that risk if you use foundation-hosted projects, but if you use open source projects, you now need to do some risk analysis of the price going from $0 to $0+n in the future.

Is this good? I mean, no? We all enjoy a social contract where we’re on the free end of getting a lot of value.

If anything, the open source world should figure out the problem of “we gotta make some money.” Historically, that’s been a difficult subject for that community to talk about, and it’s certainly never been resolved.

We also talk about a similar shift in norms when it comes to speaker phone use in public.

Take a listen to the episode!

SpringOne Live-streamed, free, next week!

Next week is SpringOne, our conference all about the Spring Framework, which I think is the most widely used Java framework. It’s free to register for and watch online.

I’ve seen a lot of the content, and helped work on some of it. You can watch it all for free next week, starting August 26th at 9am pacific time. It’ll run for three days.

If that’s your thing, sign-up for it and watch for free.

Wastebook

• In enterprise IT talk, instead of “seamless,” consider saying “easy,” “simple,” “quick(ly),” or just deleting the word altogether. (That said, “seamless” is a term of art in IT, so we sort of all know what it means based on the context of how it’s used.)

• Necessity is the inventor of compromise.

• “‘The Art of Pantomime in Church’ is this weird mashup of clowns and Christians and I have no idea how it even got made. The thesis is that professional clown Randall Bane can teach you how to incorporate pantomime into your church services. I have no dog in this fight, I’m not particularly anti-Christian or even anti-clown, but I literally do not understand how I struggle to get by and yet there was money to produce this thing.” Via.

• ”it’s a reminder that people of my generation, those who grew up watching the original (and best!) Star Wars trilogy, are now the people running the world and its weapons systems” Here.

• Commin' 'atcha with another Coté Footnote(tm).

• Incomprehensible spaghetti.

Relative to your interests

• IDC’s Worldwide AI and Generative AI Spending - “IDC has recently unveiled the latest release of its Worldwide AI and Generative AI Spending Guide, 2024 V2. Presently, the global Artificial Intelligence market stands at nearly $235 billion, with projections indicating a rise to over $631 billion by 2028.” // Something is weird with this estimate. There’s no way companies found an extra $235 billion to spent on new IT projects. And if they’re reallocating that money from other IT projects, like, your grocery buying app is going to stop working because there’s no budget for the computer needed to run it. // Meanwhile…

• C-suite enthusiasm over generative AI wanes, putting pressure on quick wins - “Interest in the [AI] technology fell 11 percentage points among senior executives and eight percentage points among board of directors, respectively, since Deloitte’s Q1 2024 survey.” // I think the people most enthusiastic about enterprise AI haven’t actually used even enough consumer AI to get a grasp on how little AI can actually do. And, as with new tech (like cloud computing in the early 2010’s), they underestimate how expensive it is to use, and how difficult it is the change how their organization functions. // Also, there’s a bit danger that the AI-driven company valuations jumps too quickly drop due to lowered expectations. // Speaking of…

• AI may be distracting organizations from other IT priorities - Well, things are getting better, but enterprise AI is a distraction from keeping the lights on. // “IBM reports that fewer than half of the respondents believe their IT organization is effective in delivering basic services, compared to 69 percent from a survey in 2013. Among chief execs, that figure is 36 percent today, down from 64 percent previously, while for chief financial officers, the figures are 50 percent now, down from 60 percent before…. What might be causing this? The Armonk-based biz says that 43 percent of surveyed tech CxOs indicated that concerns about their IT infrastructure have increased over the past six months due to the focus on optimizing their infrastructure for generative AI.”

• Private Cloud - What’s in a Name? - “Forrester’s ‘Private Cloud Market Insights, 2023’ reporting that 79% of surveyed cloud decision makers are implementing internal private clouds” (Source: "Private Cloud Market Insights, 2023," Forrester). Cf. my past attempt to round-up private cloud usage/market-sizing.

• How public intellectuals can extend their shelf lives - Spend a lot of time steering yourself to be optimistic, and engineer a continuous drip of new ideas and perspectives.

• Platform Engineering: The Next Step in Operations - This is a really good overview/state of thing. // Once again, I think it’s all about introducing product management to operations and continuously building the platform, the product.

• Developer Content That Ranks in 2024 - Educational content attracts the most attention (“how to”). Also content where you define things (“what is”). Converting that content into sales, or at least pipeline, is the secret magic.

• $ git push heroku betty - That sounds like fun. People keep telling me about the revitalization of Heroku. It’d be great to see that happen. If everything just want their own Heroku, what’s been holding them back? Solving that puzzle would be wonderful to see after all these years of PaaS yo-yo’ing. // I like the unstated opinion here: “Many are now facing lots of scattered homegrown production platforms and are starting to feel the pain of maintenance, technical debt, and difficulty in consistently delivering reliability, stability, and security while trying to scale. There simply aren’t enough engineers for every company to support this, nor does it make business sense.” // The opinion is: stop doing that.

• An American came to stay – and completely changed my attitude to water - As Europe gets hotter (not good, sure, but:), hopefully they’ll understand the pleasure of ice water.

Conferences, Events, etc.

Talks I’m giving, places I’ll be, and other plans.

This year, SpringOne is free to attend and watch online. Check out Josh’s pitch for the event. There’s an on-site conference as well at Explore if you’re interested. But, for those who can’t, now you can watch all the fun!

SpringOne/VMware Explore US, August 26–29. DevOpsDays Antwerp, 15th anniversary, speaking, September 4th-5th. SREday London 2024, speaking, September 19th to 20th. Cloud Foundry Day EU, Karlsruhe, Oct 9th. VMware Explore Barcelona, speaking, Nov 4th to 7th.

Discounts! SREDay London: 20% off with the code SRE20DAY. Cloud Foundry Day20% off with the code CFEU24VMW20.

Logoff

I won’t be at Explore and SpringOne next week. I’m hoping to write-up my thoughts and analysis from my Amsterdam perch. Of course, I know a lot of the content already, but actually seeing it presented and thinking about how I’d “live blog” it in the old days is another thing. We’ll see what happens!

//

I’m slowly starting a project called “things that are happy that you’re about to eat them.” I might call it “Please Eat Me” (maybe “Pleased to be Eaten”?) for short. I’m on the lookout out for food labels and ads that depict the thing you’re about to eat as jolly. You see this is a lot with BBQ places, like the old Bill Miller BBQ back in Texas:

Those are pretty tame. That pig seems super chill, though. You see it a lot in candy and hot dogs. The best is when a hot dog is putting ketchup or mustard on itself: getting all ready to be chomped on:

It’s a bit morbid when the food is an animal, but we should have some compassion for the fruit and vegetables too:

Also, there’s a highly-related category of, like, “Please Kill Me” (maybe “Pleased to be Killed”?) where the animal/fruit is happy that you’re buying a device that will help kill it:

If you see any, pass ‘em along!

Are you at a large organization doing platform engineering? Have you been building and/or using a platform? How are you introducing product management in your operations group?

I want to test a theory that’s come up in my conversations a lot this summer: introducing product management into ops and infrastructure organizations is too difficult. It won’t work. There are teams here and there that can do it, and they show up at conferences. But, when you're proposing that you're going to "change the culture" of thousands of large organizations, it's an impossible task. These detractors cite DevOps, even agile: after all these years, have we really done much, or have we just experienced the bad parts of Larman’s Laws?

That sentiment is pretty bleek!

If you're working in one of these large organizations, how do you start up the product management practices and roles for your platform? Is it working? A further filter is: how many apps are you running on your platform, following platform engineering practices?

I really would like to hear from you, even if it's just references to other people talking about. But, if you're working on platform engineering in a large organization, I especially want to hear from you. Hopefully, I can get enough responses to write-up how people are introducing and doing the product management part of platform engineering.

Now, here’s a long explanation of why I’d like to hear from you:

Product management is what makes platform engineering different than "what we're already doing"

I'm focusing on product management because I believe that practicing product management is what separates platform engineering from "what we're already doing." And, over the past year, this feels like it's emerged as the consensus. Once you make product management part of platform engineering, it moves the phrase past being a marketing-driven buzzword that relabels what we already have to focus people on buying new bottle for their old wine.

Product management wasn't always part of platform engineering. A few years ago when the thought leadership around platform engineering started, "platform engineering" just meant putting an internal developer portal in place (Backstage and friends). Then the platform engineering thought leadership train loaded up "making Kubernetes easier to use for developers." This caused a lot of existential angst from us DevOps and PaaS people, especially when Humanitec declared that DevOps was dead. We were all left wondering: how is this different than what we've been talking about for 15+ years? That is: what we’re already doing.

Once the 10+ year old idea of "platform as a product" was reintroduced into the platform conversation, "platform engineering" became a new enough thing that it was worthy of having its own name. It became a thing.

In my world of pre-Kubernetes PaaS, I've worked with large organizations who've been practicing platform as a product for many years, using Cloud Foundry as their platform. You could say that platform engineering is “just” a re-labeling of platform as a product, but I think it’s ended up being more than that. Platform engineering wants to do platform as a product with Kubernetes, not with existing PaaSes.1

Platform engineering focuses too much on Kubernetes

This is the second aspect of platform engineering that I think makes platform engineering a real thing: it means using Kuberntes as the basis for your platform. I don't think this is good, and I’d rather we change it so that it doesn't matter what CaaS/IaaS you use. But, that’s currently platform engineering as she is spoken.

Like I said, for 10+ years, there've been a lot of big enterprises that have used, and continue to use, Cloud Foundry to run thousands upon thousands of real-world applications. And, there's other platforms out there, not to mention all of the VM-based ways of running apps that seems to be the majority of how people run apps. There is always a platform, whether you know it or not. And if you don’t know it, it usually means its an accidental platform and hundreds of them in your organization, which is very much not good.

But, the people building and talking about platforms now want Kubernetes, it seems. They just assume there is nothing else. This has been a 7 year distraction from improving how organizations develop and run software, and a great example of how us tech people get too focused on using new and interesting tools for their own sake. And, in that way, a triumph of thought leadership and devrel.

Over those years, after finally focusing everyone on the PaaS layer, that re-focusing on the CaaS/IaaS layer has sacrificed the never ending task of improving the "business outcomes" of better developer productivity and improving all the -ility's in production. You know: becoming an "elite performer" in DORA terms.2

(Alright. I've tried to write this post a few times and it always goes dark and negative. So I'll stop it with the you kids get off my lawn existential crap.)

So, that's where we are with platform engineering: it's applying product management to building the platform, and building the platform with Kubernetes.

The two platform approaches we'll end up with

There's all sorts of people working on solving the Kubernetes problem(s): that it's complex, you don't want to expose developers to it, and you have to go to the buffet and assemble and then care for your own platform. Kubernetes is not a ready to use out of the box platform. Indeed, when you look at the CNCF cloud native platform definition, Kubernetes doesn’t even show up!

Following historic examples, the problem with building a platform from Kubernetes will resolve itself in two ways:

The Overlay: platform as abstraction layer

There will be a few winners in the "wrap Kubernetes/whatever in layer to hide it from the users" approach. This is what we're trying to do with the App Engine/Spaces framework, and, as I shallowly understand it, it's what Syntasso/Kratix is trying to do. You’re essentially saying “the APIs and config for Kubernetes aren’t tuned for the platform engineer needs, so we’re going to make the ones that are, and do all the glue work integrate that back to Kubernetes.” This is one of the most popular patterns in computering: adding an abstraction layer to make it easier to use the wrapped layer.

This platform building pattern is trying to give users (platform engineers) the ability to customize the platform to their special needs3 while still avoiding building everything from scratch, the “DIY platform.” To use PaaS-talk, this approach allows you to form your own opinions rather than (be forced to) use the opinions of your pre-built platform.

Platform engineers define the "API" of the platform components and can then build the platform out of those components. You can also throw in promise theory, contracts, and some aspects of negotiated platforms and SLOs from SRE-think. Colin Humphreys made a good pitch for this approach recently, and I’ve been dragging my feet on interviewing the Tanzu people who’re working on this approach. 

We've seen what this ends up looking, good and bad, like in historic examples, so we can start to think through some long-horizon strategy moves.

DevOps 1.0 was all about creating an abstraction layer for the mess of configuration and release management with Chef, Puppet, Ansible, Salt, etc. Years later, DevOps got bogged down in solving ops tooling and culture problems (which is great!) and rarely got up to the application developer layer - but, the intention was (usually) always there!

Then there's our current example of Kubernetes. The point of Kubernetes was to displace AWS as the standard IaaS model by creating an open standard for how infrastructure was represented, managed, and used. To use the Kubernetes term, to creat a new “API” for IaaS. It worked!

Like DevOps, Kubernetes also stalled out on the way to delivering a better developer experience. And, in fact, the Kubernetes creators eventually backed off from that ever having been a goal.

For both Kubernetes and DevOps, making the “inner loop” (if you remember that term) better was, perhaps, never the point, and thinking otherwise resulted in over-inflated expectations.

I think what the people working on the platform as abstraction layer approach want is what we hoped the Kubernetes API would be: much higher up the stack, even developer facing, and including a governance framework. Essentially, a developer-ready system with all the enterprise grade blah-blah tools.

This is great! It'll be fantastic if it works!

What’s important is that you have to product manage all of this to figuring out (1) what those overlays are, (2) how you customize the overlays for your environment, (3) how you pick and choose which overlays to assemble into a platform, and (4) when you add in new functionality.

The vendor and open source community around the overlay will help with some of that, especially number one and four. The vendor/community will gladly tell you what it thinks the defaults should be and even provide you an out of the box, ready to use enterprise grade blah blah platform (see the next section) based on those overlays.

But, the whole point of the platform as abstraction layer is customizing the platform to the user’s needs, so, like, the user needs to do that. And product management is how you do that.

Paas is cool again

Since 2007 we've been through several cycles of people trying to build their own Heroku.

It goes like this:

1. A new platform comes out that makes it easy for developers to build their app, connect the app to a database, etc., deploy the app to production, and scale it for performance needs all on their own, self-service, using the latest frameworks and services.

2. It only runs in the public cloud.

3. Large organizations initially reject it for two reasons: (a) it actually will not scale. (b) it needs to be on-premises for very important enterprise reasons. "We want that awesome developer experience and velocity," the large organizations say, "but, uh," looks down at notes, "I was told that we work in a 'highly regulated industry.'"

4. This brings in phase two of the cycle: vendors make on-premises platforms. Strangely, maybe even heroically, Heroku never entered this phase. But, you saw it with things like the container wars of the 2010's, which drove on-premises platforms like Cloud Foundry.

5. Our attention goes back down the stack to infrastructure instead of the fully build PaaS. PaaS is no longer cool. The last phase of the cycle is usually caused by a disruptive technology coming along and pulling the user and buyer's attention away form the now boring platform that just works. Docker was the first to disrupt PaaSes in the late 2010s, and then Kubernetes came in. In both cases, the user assumptions was that each was a viable PaaS replacement. But, as capital-D Disruption, in each case, neither was a full-on replacement for all the enterprise grade blah blah. And yet, these less feature-ful disruptors drove organizations away from the PaaSes that they once loved. E.g., Heroku and its children.

I've made a satire out of that cycle and the thousands of highly paid professionals who make those decisions along the way.

But, I mean: I'm not sure why people don't just use Heroku or its many on-premises focused descendants like Cloud Foundry.

There's "fashion." People wanting to use something new. I've heard this sentiment many times recently: "Our PaaS is great, but if I don't learn Kubernetes, it'll be harder to find a new job. So, we need to migrate to Kubernetes." What with enterprise Kubernetes build -out being just at the beginning, that sentiment is optimized for an individual, but not for the organization that already has something that works in place.

Other than "fashion," I think what drives people away from PaaSes is: 

1. Pricing - The first grumblings about Heroku were that it was awesome, but expensive. This has been a common sentiment about all PaaSes. Showing the value of any platform is hard, so it always looks expensive when you start running thousands of applications. All you see is a price, and linking it to the revenue that those thousands of apps drive is not a normal way for traditional organizations to think. They treat IT as a cost center, not a part of the business. So when someone comes and says "we could build that platform ourselves and remove however many millions in licensing/cloud cost," management gets excited.

2. Difficulty in customizing or "swapping out" components, and lack of new features. The last is what introducing product management into platforms is trying to solve: you're actually supposed to talk with developers and deliver new platform features that solve their problems and make them happier.

In this part of the cycle, the Kubernetes problem is solved by hiding or even removing Kubernetes. It doesn’t matter what CaaS/IaaS is at the bottom of the PaaS. Maybe there’s even two, or three! You might even introduce a totally a new compute, uh, “paradigm.” Maybe serverless will finally fulfill those Wardley-dreams! Unikernal, WASM - whatever! Just a few weeks ago my mind was thoroughly exfoliated in the sauna with the idea of Isolate Cloud. ANYTHING COULD HAPPEN.

The role of product management here is different. If you’re using a PaaS, you’re not given a lot of tools to do all of your own customization. You more rely on the PaaS vendor and community to do much of that work. The overall community does most of the product managering, which mans they need to do a lot of it. It also means you need to upgrade the PaaS when there are new versions. That is a big challenge. People at large organizations don’t like upgrading their stacks.

Maybe one way of looking at it is that a PaaS outsources platform product management. It’s probably more that what you’re product managing is (like the platform as abstraction layer) the selection and assembly of pre-built components.

Without product management, platform engineering will struggle

This is why I've typed this far: the long-term success of platforms relies on product management. Once you stop adding new features to the platform, people look for new options. If you can't customize it to your needs (real or just made up enterprise blah blah), you'll start the platform cycle all over again. This means you don't get the full benefits of many years of platforming. You'll start neglecting what you have, focus on migrating your existing apps to new ones, and experience those ups and downs of benefits achieved when you see surveys of new tech usage. ROI depends on years of payback, and if you restart in the middle of that period, the math doesn't work.

So! That's what’s driving my interest in learning how organizations are introducing and sustaining product management in their platform groups.

How's it going for you?

Logoff

Colophon

Three things go me thinking on this above, which I want to call out:

First, this from Figma’s write-up of migrating to Kubernetes:

Having users define services directly in YAML can be confusing. Instead, we worked to define a golden path for users and allow customization for special cases. By being explicit about what users can and should customize—and otherwise enforcing consistency by default—you’ll save users time and energy while also simplifying maintenance and future changes.

That seems like a pretty compact explanation of the developer-facing goals of platform engineering.

Second, as referenced many times above, Colin’s excellent piece of “let’s make better mistakes this time” for platform engineering enthusiasts. I like that he managed to sneak in a link to an old version of Pivotal paper on platform engineering.

Third, Betty announced that she’s the new CMO at Heroku. When I was looking around at new job opportunities this summer, almost every person told me I should talk to the new people at Heroku. They’re trying to get the band back together, etc. I don’t really know anyone there, and I decided that my current job is just fine. But, it’s exciting to think that Heroku will start doing more in the platform area. I think the timing is right for a public PaaS only play - you’d be cutting out a lot of enterprise customers, but over the next five years, I think all that enterprise blah blah will be less of a barrier. I mean, that’s what all the CIOs are saying: in three years they’re planning to almost double the amount of workloads in public cloud. Maybe they’ll actually do it this time.

While writing this, I came across this huge write-up of platform engineering ahead of releasing a new book on the topic. It’s great! I’m looking forward to that book.

Also, we have our big annual conference next week. I’m thinking we’ll have some options about the above. You can watch some of it as a live stream, or scrounge for coverage and videos later. I’ll write-up anything relevant here. Probably.

There’s something I chopped out of my initial, thankfully thrown away draft of all this: we don’t really talk about “operator experience” in platform engineering. It’s very developer focused. This is fine! But, more than likely (as with DevOps), the story will turn inward to making the tools and lives of platform engineers better. Put another way, we’ve long had the 12 factor app manifesto, but what is the other side of that, the 12 factor platform manifesto?

Figuring out what kind of podcast you’re doing

One of my co-workers is started a podcast and had some questions. Here’s my answers. As with most “how do I do this?” sessions, it focuses a lot on gear which I scoot away from in my answers, you know, following the cliche that the tools matter less than what you do with them. Also, the tools are tedious, but easy to learn.

What podcasting software do you use?

This is all for recording remotely, over the Internet. For recording in person, there’s a different set of gear and practices.

For SoftwareDefinedTalk.com we use Restream. When I’m recording an interview, I use SquadCast which is now bundled in Descript subscriptions.

I suggest using one of those services so that you can record separate audio tracks and get video (whether you want the video or not). You don’t need to livestream your recordings.

Before those services existed, I’d use Audio Hijack Pro to record the incoming audio and outgoing (my) audio. But that’s a lot of work compared to just using one of the above. You can also use Zoom. I think it’s pretty good, actually, but I haven’t checked it out recently.

What are the features of that software?

Recording.

The main thing each does is get a video chat going and record separate audio tracks. Most of them, now, will record the audio locally (on your and your guests laptop) as well. So, if the connection is bad, you’ll get good audio in the final recording. These services usually will give you recordings of each participant and a recording with them all together.

Editing

For Software Defined Talk used Hindenburg for many years (ten?!). It’s at the sweet spot between lots of functionality and easy to learn and use. It’s built around editing voice, has good, basic audio filters and effects (leveling, noise reduction, etc.), and good setting for exporting podcast MP3s (setting cover art, putting in chapters, etc.) It looks like it had editing by transcript.

Now-a-days when I edit podcasts I do it in Descript. You can edit the video (or just audio) and export an audio only MP3. It has all of the export things that Hindenburg does.

If you want a free option, Audacity is probably pretty good. That’s what I used long ago in the 2000s and it was just fine.

Distributing

Here, I assume you mean hosting a podcast: uploading it somewhere so that it makes an RSS feed that people can subscribe to.

In addition to hosting the MP3 and creating the RSS feed, the service should create a basic podcast website for you. You want an index page that lists each episode and then each episode should have its own page, the show notes. You want to be able to edit those show notes to put whatever you want in there. The service should allow you to automate all of this each time you upload an episode.

Each episode should have a URL you can go to. The better services will automatically create a URL based on the episode number, for example: https://www.softwaredefinedtalk.com/480.

These services should also have metrics too for episode downloads. You can’t tell much from podcast downloads: just number of downloads and geography. Some of the nicer services will pull in more detailed metrics from Spotify and Apple Podcasts.

I’d also look for a platform that manages your podcast listing in Spotify and Apple Podcasts. You can actually do that on your own, but many of the services will do that for you.

We use fireside.fm. My co-workers use podbean.com and transistor.fm. I haven’t used it in a long time, but libsyn.com is one of the original hosters and probably good.

A lot of big name, highly professional podcasts (the NPR style below) have terrible show notes. It can be hard to even find where the show notes are. This is baffling. Have show notes and good ones. Check out the example podcasts I cite below, they all have good show note sites…except the most professional one! Here is the test: if a person wanted to link to an episode, is it easy for them to find the web page (does it even exist?!) and get a URL that they can share?

Do you edit your podcasts? If so, any advice?

I don’t edit the Software Defined Talk ones anymore, one of my co-hosts (Brandon) does. But I used to for 10+ years. Plus, I edit the Tanzu ones and one-off ones I do. Brandon is a much better editor than I am.

Learning the basics of editing take a little bit of time, but it’s basically chopping up your audio file into chunks, arranging the chunks in the order you want, and deleting the chunks you don’t. For most podcasts that are interviews, you won’t be re-ordering the chunks. Instead, you’ll delete parts you don’t want. Most podcast editing is that: deleting parts you don’t want.

You’ll know you’ve gotten good at podcasting editing when you start to use the keyboard short cuts, and when you can imagine the edits you’d make as your talking in the podcast. As you listen to other podcasts, you’ll also be able to detect when they’ve done edits.

You can put music and standard bumpers in the intro and outro. I’m not a huge fan this as a listener or maker. If you do have music, make it brief and use the same music each time. This will help listeners slide into the comfortable sameness of each episode.

Do you have any general podcasting advice?

I would decide on the style of show you want and then edit accordingly.

Do you want (1) the full This American Life/NPR style podcast with summing up intro and hook, music in the background (“ducked”), narrator spliced in with interviewee comments, edited segments, and music? Or do you want (2) the cold open, we just recorded two people talking style? There’s an in-between that most people do where (3) they have some intro music that fades out, and maybe they do some editing of segments.

I do the second type because as a listener I don’t like the others and it’s the quickest and easiest to make.

Deciding the format and style you want is important. Once you have that you can use it to guide what you do like editing and even selecting guests. The style you choose will determine how you structure and run the podcast, and also what you’ll do during it.

If you’re doing the NPR-style, your goal is to get other people to do all the talking and edit together a story. NPR-style podcasts aren’t really podcasts at all: they’re professional radio shows that are distributed as podcasts.

If you’re doing the other two (“real podcasts”), you have to remember that listeners are as interested in you as they are in the guests and topics.

The success of “real” podcasts relies on establishing a parasocial relationships with listeners, often lasting many years. Listners want to hear the topics, learn abut things, but they also establish a friendship with the hosts. Listeners look forward to hearing what their friends are up to each week - hanging out with them each week.

This means that if you’re doing this style, if you’re interviewing someone you should interject what you think instead of just asking the guest questions. And if you’re discussing some topic, you should offer lots of opinions and “thinking outloud,” not just covering “the facts.”

Also, this means you should be consistent in your release period (weekly, every two weeks, monthly) and post it on the same day. For example, we post Software Defined Talk episode every Friday at 7:30am Amsterdam time. I know the release days for all of my favorite podcasts and look forward to them.

Here’s another distinction in types of podcasts: guests and no guests.

Having guests is having (mostly) different people on each episode to interview them about whatever - in tech, usually something they’re an expert, or at least knowledgeable on.

In these cases, you want to get the guest talking most of the time, but you should also think of yourself as a proxy for the audience to ask and clarify things that the guest is saying. It’s also good to rephrase things from time-to-time, and then, as always, offer up your own thoughts.

People have a lot of opinions about length. You should ignore that advice and figure out what you like and what works for your style. I like long podcasts, some other people like short ones. If someone tells you to target ten minutes because “people don’t listen to long podcasts,” chances are this person doesn’t listen to podcasts. This comes up a lot when you’re doing an official company podcast. Only take advice about podcast length from people who actually listen to a lot of podcasts.

There is no magic number for length. Start with having it be slightly longer than it needs to be, and back up from there if you want to.

Keep in mind that people won’t always listen to a podcast beginning to end in one sitting. They might pause it and come back to it. Sometimes it takes me two or more days to fully listen to a podcast.

Again, your listeners are there to hang-out with you each week - they’ll appreciate you spending more time with them rather than less. And, if it takes them several days to listen to it, they will because they’re interested in hanging out with you.

(4 hours might be too long, but I’ve listened to many enjoyable 3 hour podcasts over the past ₂₀ years, and hundreds of good 2 hour ones.)

As with writing, the other thing you should do is listen to a lot of other podcasts in the style you like. They can be in your topic area or a completely different one. For example, even if you’re doing a tech podcast, you can learn a lot about podcasting from The Flop House and Blank Check.

Once you’ve made three or so podcasts and edited them, you’ll be able to pick apart how good podcasters operate and start learning from them. And, also, you’ll just learn the format, and what you like. With this you can figure out the podcasting style you like and start to build - and refine/evolve - your style from that.

Here’s some more podcast example:

• The Political Gabfest - this is a very professionally done “the same people three people talk about the news of the week” podcast. You can see that they stick to a format - there’s usually three topics, the host intros/summarizes the topic and then asks the other two hosts what they think; the host will say what they think to. Also, at the end of each episode each host recommends something. I stole that idea for Software Defined Talk when we started our podcast. This is the podcast that has terrible show notes - see how frustrating it is the look at their page?

• Software Defined Talk - this my podcast, so it’s obviously the best in the world. We patterned it off The Political Gabfest, but after 10 years it’s developed its own style. I would call this style “the mess that works.” It’s the same three people each week discussing a selection of tech news and any side topics. Here, the format is: (1) Coté’s inane cold open, (2) one to two tech topics we discuss, (3) listener feedback and interesting conferences, (4) recommendations from each host, (5) usually an after show with some goofy outtakes or an extended, off-topic conversation. A more professional version of this is Sharp Tech, and the Cloudcast News of the Month.

• Oxide and Friends - this is another “mess that works” example. It is very much a personality-based podcast, based on Bryan Cantrill. The format here is the classic: his co-host Adam is the straight man and Bryan is the goofball. This is also an interesting podcast to study if you’re making tech podcasts and you’re making an official company podcast. Oxide is on an impossible mission: making on-premises hardware! So, part of their marketing is embracing the absurdity of that business scheme and somehow converting it to tech-cult energy, to hope and belief that on-premises hardware is cool and interesting. It’s like the early days of Kubernetes where Kelsey would run around doing demos. This podcast does a lot more than that, of course, but part of it is feeding that tech-cult energy.

• The Changelog - I don’t listen to this podcast much (it’s usually too technical for me), but it’s very popular. Compared to other podcasts it’s high-production.

Finally, with most of my advice about content, my number one recommendation is to just do it and publish. The moment you hesitate because you think you should edit more, polish it up, or (worst) re-do it, you should instantly click publish.

If you’re doing this right, you’ll always think the episode could be better, you might even think the episode is bad. You might even regret questions you asked, that it’s boring. Just keep publishing. It’ll take time to discover your style. Eventually, the audience will find you. But if you engineer it too much you get the equivalent of “data sheet” PDFs and shallow enterprise software web pages. Something that looks perfect, but says nothing.

Publishing the imperfect is how you get around that. As I keep emphasizing, a “real” podcast is about establishing a very intimate relationship between the hosts and listeners.

By doing it weekly and publishing it you start learning how to choreograph that all.

We plan on having guests on our show… what do you do to help guests have the best possible audio quality, especially if they’re not likely to have professional equipment?

Getting good audio from guests can be tough. The first thing is to get the to use headphones. Then having them use something other than a laptop mic. If they have a headset with a mic, that’s good. And iPhone headphones with a mic are actually pretty good; you just have to ask them to hold the mic away from their beard if they have a beard.

Very importantly: make sure they check their audio settings to make sure they’re using the good mic. If their audio sounds off, ask them to double check. I’ve been doing podcasts for about 20 years, and I mess this up as a host and guest at least once a year.

Also, experiment with using the Continuity Camera with a iPhones. The mic can be pretty good, and the placement of the phone mic on the top of their laptop screen will position the mic well, pointed at their face.

And beyond audio quality, you need to learn how to steer your guests. This is one of the things you’ll learn from listening to other podcasts. The best steerer is Tyler Cowan, though, I could see how some people would find his style insulting and cringe-y: too controlling.

And while not a guest, if you’re doing the multi-host thing, you need to learn how to be the host that runs the show and moderates it (let’s call it the active-host) but also how to be the other hosts (the passive hosts).

For the active host, on our podcast, Brandon is good at this (which happens in the episodes I’m not on), and Plotz is good at this in The Political Gabfest.

The important thing about being a passive host is that you have to stop talking when the host changes the topic. You’ll get tuned to each other, and even pick up on the natural rhythms of each other. You’ll know about ten seconds ahead of time when a person is wrapping up so you can prepare to go next.

A lot of the convention about how to be a “good listener” in life doesn’t apply to podcasting. You have unlearn polite conversation topics, and almost do all those things that people tell you are offensive and rude in meatspace talk. Interrupt each other, talk over each other, say what you think. Or not: it depends on the style you’re doing. But don’t bring all the “how to be a good listeners” and nonviolent communication stuff to the table right away: evolve to that if that ends up being your style.

In summary, what I’m saying is: finding and tuning your style of podcasting is more important than gear. Part of that is creating the your podcasting character and being conscious of playing that character. It might be a lot different than the character you are in other parts of your life. You listeners want to hear that character, they’ll become friends with it. Your listeners won’t show up just for that character: they want the actual interesting content and guests! But, that character is what will make your listeners keep listening, subscribe, and fill in the gaps between impossible awesome and what you actually published. So, you have to cultivate that character.

Relative to your interests

• Why Every Java Developer Should Attend SpringOne - the online version is free to watch on August 26th to 28th.

• Travel Spending On Track To Return To Pre-Pandemic Levels By End Of 2024 - ”Global travel spending is roaring back and will fully recover to pre-pandemic levels by the end of 2024, surpassing $2 trillion.”

• Walmart used AI to crunch 850M product data points and improve CX - “The primary in-store improvement is that associates’ can use in-store technology like mobile tools to quickly locate inventory and get items on shelves or to waiting customers — a significant upgrade from the ‘treasure hunt’ of finding items in years past” A few more details here.

• Apax to take consultancy firm Thoughtworks private in $1.75 bln deal - Summary from Justin Warren: “Software consulting company Thoughtworks is getting bought by private equity firm Apax Partners for about $1.75 billion. Thoughtworks has been struggling a bit lately, with revenue down 12.4% YoY to $251.7 million and the stock down 87% since 2022. It’s already done a bunch of cost-reduction restructuring, and it looks like more is on the way with 630+ staff to go” out of ₁₀,500 staff.

Wastebook

• “billboards festooned with three-dimensional meat bearing messages such as ‘You never sausage a place!’” Here.

• It’s hard to forget a bad idea, and you have to work hard to remember good ones.

Conferences, Events, etc.

Talks I’m giving, places I’ll be, and other plans.

This year, SpringOne is free to attend and watch online. Check out Josh’s pitch for the event. There’s an on-site conference as well at Explore if you’re interested. But, for those who can’t, now you can watch all the fun!

SpringOne/VMware Explore US, August 26–29. DevOpsDays Antwerp, 15th anniversary, speaking, September 4th-5th. SREday London 2024, speaking, September 19th to 20th. Cloud Foundry Day EU, Karlsruhe, Oct 9th. VMware Explore Barcelona, speaking, Nov 4th to 7th.

Discounts! SREDay London: 20% off with the code SRE20DAY. Cloud Foundry Day 20% off with the code CFEU24VMW20.

Logoff

I’ve spent a lot of time “researching” and learning this week, not much producing content. It’s like input versus output. This is difficult, mentally for me. I only value publishing, so if I’m not doing it, I don’t think I’ve done much work. This is not healthy!

To get around that by thinking about what other creators do. How often do they publish? How much time do they spend “puttering around” building up to the ideas they eventually publish. How much stuff do they work on that never gets published.

When I’m in this learning mode, I’m creating lots of content: I’m taking notes, trying to rewrite things. I’m also talking with other people a lot, testing out my understanding of the “input,” getting them to explain it to me, thinking about things we could do as output. It’s sort of like a podcast that never gets published.

//

Part of that “input mode” is listening to lots of videos. I can’t watch videos at my desk. I’ll get distracted with other work and then I’m no longer paying attention. This a good opportunity to walk the dog on long dog walks. So, this week, I’ve found several new Garbage Chairs of Amsterdam.

Just fun finds and links today.

Wastebook

• “FWD: RE: radioactive fungus email from grandma” Here.

• “I don’t know about you, but I think a campaign setting ruled by evil angels and their witch-wives, populated by giants (perhaps not 3,500 metres tall) who eat one another and human beings, and who have sex with animals to produce many weird varieties of beastman, is one that somebody could do a lot with.” Here.

• ”AI enables action without thinking.” Here.

• “Agent Double-O Soul, baby.” Edwin Starr.

Relevant to your interests

• How We Migrated onto K8s in Less Than 12 months - Always hide the yaml: “Having users define services directly in YAML can be confusing. Instead, we worked to define a golden path for users and allow customization for special cases. By being explicit about what users can and should customize—and otherwise enforcing consistency by default—you’ll save users time and energy while also simplifying maintenance and future changes.”

• 5 Lessons For Building a Platform as a Product - They’re doing a good job trying to evolve the Pivotal Cloud Foundry philosophy of platforms. // “I talked to a CTO at one of the world’s top banks, who explained that he loved what Cloud Foundry could do but wondered what would work for the other 99% of workloads he had responsibility for.”

• Who uses LLM prompt injection attacks? Job seekers, trolls - ‘“At present,” Kaspersky concludes, “this threat is largely theoretical due to the limited capabilities of existing LLM systems.”’

• AI or bust? Only one part of US tech economy growing - “Assuming the bubble does not burst, S&P forecasts global AI spending to grow by more than 20 percent through 2028, when it is estimated to account for 14 percent of total global IT spending, up from 6 percent in 2023.”

• How to go-to-market: Measuring Marketing Value - “The key areas that your marketing team can drive impact for the business are in Awareness, Engagement, and Pipeline.”

Logoff

Some local under-the-bridge graffiti: