I’m slowly working through a new edition of my Cloud Native Journey booklet. Here’s a draft of the chapter on compliance, auditors, and all that. The goal is to draw on actual cases and lessons learned from organizations.
In which I try to figure out organizations should do with kubernetes, let alone what the thing is.
There’s no recording of this talk yet, but here are the slides.
In the cloud, DevOps, agile, whatever is hot and new era, the role of enterprise architects is rarely addressed. There’s probably plenty useful for them to do still. I’ve been trying to figure out what those things are recently, and here’s my trip report, an older version:
Also, see the slides, which are usually more up-to-date than that older recording.
Here’s the slides for my Creating Better Software presentation.
I usually use variations of these slides for two of my talks: “Not actually a DevOps talk/Creating better software” and “Pragmatic digital transformation, or rescuing your business from decades of IT quicksand.”
In a mildly surprising announcement (see below), Docker announced a new CEO, Steve Singh, formerly of Concur-cum-SAP. The consensus is that this is the typical startup move to get a “more enteprise-y” leader in place, e.g., this happened at Puppet, Chef, and probably MuleSoft and Cloudera [I don’t know those last two well enough, but ¯_(ツ)_/¯.
As most all non-Docker (myself included and us over on the podcast) are always saying, Docker, Inc. needs to figure out a business model. So huzzah on that!
The PR folks at Docker were nice enough to line up a bunch of interviews. Below are some highlights from them and other places, As always, my comments are in square brackets unless it’s obvious otherwise.
From outgoing CEO Ben Golub’s blog post announcing this:
- Customers, in addition to the above: “Docker has rapidly scaled revenues, building a sustainable and exciting subscription business in conjunction with tens of thousands of small and mid sized businesses and over 400 G2000 customers like ADP, the Department of Defense, GE, Goldman Sachs, Merck, MetLife, and Visa.”
- Partners: “we’ve created enduring partnerships with the likes of Accenture, Alibaba, Avanade, AWS, Booz Allen, Cisco, Google, HPE, IBM, Microsoft, Oracle, and more.”
Singh Interview with Todd Bishop & Tom Krazit
- Over at Geekwire.
- “Today we have about 400 enterprise customers, and that’s really been over the course of the last year or so that those enterprise customers have adopted Docker, so the rate of growth is phenomenal. I think our capacity to add value to those enterprise customers is huge. They’re running legacy applications. They’re building new applications on top of the Docker platform, and that’s driving massive economic savings for them. They’re able to see 75-500 percent reduction in the application infrastructure.”
- It’s rare to see cost reduction used as an argument by cloud-native vendors; usually it’s a value-based sales: “compared to all the growth-money your company will make with this (and, sure, productivity gains), the cost will be nothing.” This might be an early sign of Docker’s enterprise value-prop: it costs less. One part to ponder: how cheaply can they do this w/r/t to sales and marketing? They have some good marketing costs from brand/community – i.e., everyone knows Docker. They need an enterprise salesforce and sales engineers to help buyers evaluate and then procure stacks; that is, unless they go (mostly?) pure inside sales a la Solarwinds. There’s some interesting comparisons here: Red Hat with the pure open source play, traditional middleware & infrastructure vendors with the open-core model, and then Solarwinds and New Relic for more inside, service-driven sales.
- “My last day at SAP was April 30.” [So two reasons to wait until now (vs. at DockerCon): (1.) it’s traditional to let a standing exec have “one last conference,” and, (2.) Singh still had to officially be out at SAP.]
- Some product-think, CaaS and building a platform with container parts: “We’re in the fortunate position that we are the leading platform company in this broader container-as-a-service space, number one. Number two, if you think about orchestration — things like Swarm or things like Kubernetes — our view is that what our customers are looking for is a platform to run their applications, manage them and deliver them. We want to take an open platform approach that allows the customers the choice to pick whatever components that they want from Docker and whatever components they want from anyone else, and run them in a model that makes sense for them. We have a number of customers that use the Docker platform as well as components from either the open source community or partners. That’s fantastic. When customers have choice, that’s a great thing.”
- He skirts naming any competitors, which is standard, but, you know, eye-rolly.
Interview with Ari Levy
- Over at CNBC.
- “The software start-up was generating less than $5 million in revenue [in 2015].”
- “Singh inherits a business that’s growing, with annual revenue in tens of million of dollars, according to sources with knowledge of the company but who asked not to be named because the financials are private. He’s still got money in the bank from the $95 million round that Goldman Sachs led two years ago.”
- I’d read “tens of millions” as $20 to $30m. If it was $50m, they’d probably say so; if it was $40m, they’d say “nearing $50m”; and if it was $10m, it wouldn’t plural. Also, it doesn’t specify if this is TCV (paid out over 2-3 year contract terms) or ACV (total size of the deal, over all years of the deal’s life-span), but, hey, that’s how all private companies speak of revenue now-a-days. As a point of reference, the product my company, Pivotal, sells in this space: “booked revenue grew 130% (in 2016) to $270 million from $117.4 million for 2015.”
- “The company started selling its enterprise product early last year and has about 400 customers, Singh said.”
“I’m more focused on the road ahead as opposed to things that have happened in the past,” Singh said. “We want to collaborate and make sure customers have choice.”
- Scott M. Fulton over on The New Stack has an interview. He tries the old IPO question, resulting in a predicable non-answer.
- If you can bare it, there’s some interesting sentiment over in the Hackernews thread.
- RedMonk’s Fintan Ryan pokes at some shifting Docker user/customer demographics in his DockerCon debriefing, e.g.: “we commented during the event that the average age seemed to have gone up by five years) and overall a more enterprise feeling at the event”
- For more context, see my recent round-up of container surveys, which also has pointers to TAM estimates.
- We’ll likely discuss this in today’s Software Defined Talk recording. Subscribe to the podcast to get it!
My The Register column this month is on scaling DevOps/cloud-native teams to the entire organization. It’s easy to build one team that does software in a new and exciting way, but how can you move to two teams, five teams, and then 100’s? It goes over the amalgamation of a few case studies and plenty of over-the-top gonzo analogies, per usual.
Over the decades, the number of H-1B workers allowed into the US each year has grown. With the 1998 update, the visa cap lifted to 115,000. In 2000, the limit was boosted again, this time up to 195,000. That year, the law was also tweaked so that renewals no longer counted toward the cap. In 2004, the cap was reset to 65,000, but an exemption was added for 20,000 students graduating from US institutions with master’s degrees. Exemptions were also added for workers affiliated with academic institutions, which can include schools and teaching hospitals. According to Ron Hira, a professor of Public Policy at Howard University who has studied the H-1B issue and testified about it before the Senate, the actual number of visas handed out each year has been around 135,000 over the last five years. Link
There’s a good rant on the relative importance of all of this in last week’s Political Gabfest. While us “on-shore” workers in the tech industry may see that 135,000 as a threat to our cashflow, it’s a drop in the bucket of employment in America. As Adam Davidson argues well, therefore, worrying about H1-B visas should be pretty low on the list of how to setup up more people with good jobs:
The question of H-1B visas has rhetorical importance far beyond its actual economic relevance. The unemployment rate for computer and mathematical occupations is, currently, 2.1 per cent. This is what economists consider full employment, meaning that pretty much everyone who wants a job has a job or is in a brief hiatus between positions. The number of jobs in those fields is growing fast—by about twelve per cent a year—and the number of qualified workers is not growing enough to catch up. In short, the plight of computer professionals is on few people’s list of urgent concerns…. According to the Bureau of Labor Statistics, ten thousand computer professionals start a new job every working day. In this context, the eighty-five thousand foreigners given H-1B visas each year represent little more than statistical noise.
He goes off an a political jag after this saying that the H1-B discussion is a proxy for “fear of brown people,” which certainly has appeal to leftist people like myself. There’s a business question here, too, though: are H1-B visas a good idea and why? Are they ethical and effective?
What types of jobs?
Also, some interesting analysis of the types of jobs H1-B visas are used for. Mostly for jobs at outsourcing firms:
But it’s how H1-B visas are being used by applicants that’s really changed. Data from the 2016 batch of H-1B petitions show that the top 10 sponsors of H-1B visa workers in the US are all corporations with large outsourcing businesses: Indian companies like Infosys, Tata, and Wipro, which pioneered the business, and US-based firms like IBM, Accenture, and Cognizant, which saw the success of the Indian contractors and began offing their own competing outsourcing programs. Those 10 firms have more workers currently employed through the program than the next 90 companies combined, a group that includes all of America’s largest tech companies and banks.
So, the discussion about H1-B visas in tech ii, by bulk, about the 60,000+ jobs in IT outsourcing. This is in addition to the estimated 1.7m off-shore jobs in outsourcing already.
In theory, most of these are “lower value” jobs where you’re more operating IT (help-desks, managing the daily operations of enterprise applications) rather than creating it (like programming). Anecdotally, there’s still programming running around in there, esp. when it comes to modernizing applications. The going theory is that you can’t just slot in workers on higher-value IT work like writing custom software.
How do you think about all this?
There’s an odd ethical vs. business-sense argument scurrying about as well that I’ve never seen addressed. One, you’d seem to be happy that the H1-B visa worker was getting work. By nature of accepting the job and up-rooting themselves, it must be good for them: or, at least, better than other alternatives. Also, if it’s actually cheaper to get the same services/output from an H1-B visa worker, why would you pay more for “native” worker? On the other hand, it’s equally confusing to figure out what companies “owe” workers that they’re firing in favor of the H1-B visa workers.
Tech companies like to skirt all that by talking about “we have to hire from a global pool,” which is fine if you’re hiring for an individual with unique skills. However, the divide between outsourcing firms and tech companies suggests that the bulk of H1-B visa hires in tech are not for all the super unique, AI experts that may not live on-shore. Then again, it’s insulting to even think that: why do I value one set of people over another in any context?
Businesses say they’re not satisfied
However we figure out talking about this, it’s clear from surveys that companies are dodgy on the value of outsourcing. As I summarizing some HfS work recently:
Outsourcers too often do exactly what the contract (from five to ten years ago) says instead of helping you innovate and keep the business growing. Itʼs little wonder that in a recent study, more than 75% of senior executives said they want to replace their legacy outsourcers because those providers are so unwilling to change to new models.
If we take Adam Davidson’s perspective, it’s not really even a problem worth thinking about (versus all the other hair-on-fire issue we have). However, when it comes to outsourcing (which I’ve shifted to because so many H1-B visa workers end up at outsourcing firms), it’s clear that we could be doing much better.
Long ago, Spanning Sync was the only viable way to synchronize your GMail calendar and contacts with the (then) OS X iCal and Address Book. It was great! I also know one of the original founders, Charlie Wood, and we’d talk from time to time about the growing company. At some point, it became a Google Apps (now “G Suit”) back-up service that had a clever value prop: cloud storage, sure, but it’s not redundant you know, you gotta do the basics.
Anyhow, I always kept a close eye on the company. It was a little odd to see EMC buy them back in 2014: as VMware demonstrated with their dropbox competition products years ago, Apple is pretty goofy here, and even Google has demonstrated over the year, large software companies are pretty and at long-term plays for individual software; Microsoft is of course an exception with Office and sort of proves the rule.
We’ll see what Insight Venture Partners does with them. I’m guessing if you just left Spanning alone, more or less, it’d turn into a cash machine at some point. That said, I don’t think Dropbox and Box are exactly profitable. Here’s Box’s last four financial years:
…but it seems like a back-up service could controls costs better and do a lot less marketing: Box and Dropbox have been acquiring companies and re-positioning themselves as they go from more than just to cloud storage to something like “sort of Office, but not really, but maybe – or like Trello… er… let’s acquire another company and go to a conference where we have wooden floors and free espresso in the booth and think about this at next year’s company retreat in Italy.” (I KID! I KID!)
Here’s some Spanning momentum from one of the write-ups:
Spanning has seen 70 percent year-over-year revenue growth and more than 7,000 customers, according to a press release. It restored around 18 million items for customers in 2016, and expects to continue growth with its global data center expansion, and distribution agreements with major channel partners.
A wet-finger-in-the-wind business case
It’s hard to quickly find pricing for Spanning on their page (smells like enterprise software!), but a few searches, particularly from Spiceworks, says it’s like $35 a month.
There’s certainly discounts on some of those customers, but let’s say the revenue would be a max of $2,940,000 annual to something like $1.5m on the low-end if you do all sorts of discounting on clusters of users.
Now, 70% y/y growth is pretty impressive, but not too insane for a relativly new offering. Let’s say they do that two more years and then it goes down to like 30 or 40% for any length of out years we care about.
Then, let’s just take a swag at storage costs. Who knows if they use S3, but let’s assume they can get down to similar pricing, we’ll take S3’s mid-tier: $0.0125/GB/month. My work Google Drive says it’s 22 GB, but I save a lot more stuff than most people do. Let’s just go with 20GB as an average. Then let’s assume you at least duplicate it, so you’re paying for 40 GB a month (across two cloud zones), which is $6/year. (Let’s ignore networking transfer charges – adding that in is left as a exercise for the reader!)
Then you need all the meat-sacks. You could probably get by with 6 to 12 product staff (programmers, product manager – you probably outsource design at this point as needed).
You need the CEO, HR, CFO, and probably 1-2 people to work for them (6 people max); you could probably cut out HR depending on how Insight likes to run HR (outsourced or pooled across companies). Maybe the CFO, but probably not.
I’m no enterprise SaaS business expert, but I’m guessing it’s marketing and sales heavy, so:
Then you need probably 2-3 people in marketing (if you were slick, you could outsource a lot of this, esp. for something as easy to understand as “backup”): 5-10 face-to-face enterprise hustlers, and let’s say a team of 5 “inside/web” sales people who send all those annoying “Re: catching-up. I see you read out white paper on BACKUP. Would you like to talk more? Are you the right person at your organization?” emails. So, max 18.
That’s around 36 people, which seems really low to me. But, if you were, I don’t know, a private equity firm, you’d probably think that was OK, if not a little heavy for a company that basically just copies files from one place to the other (yes, I’m being MBA-fatuous).
Without getting a spreadsheet to do some clustering, doing salary cost across such a diverse set is hard. Many of them are in Austin (I assume, still), so let’s just of with $150,00 all-in per head (I’m sure the admin staff and your “strategic account” sales people get paid well plus extra comp, and the more senior tech staff get paid more). So, that’s something like $5,400,000 in people expenses. Then there’s going to conferences, probably a large ad budget, that nice office they have in downtown Austin (which I think is an EMC office, so they’ll get the boot?) which means buying a lot of organic beef-jerky and craft beer etc., then there’s flying those 5-10 enterprise hustlers around and their $70-100 a day per diems, plus wining and dining. Let’s just trow in another million and go to $6.5m.
So, with some mumbo jumbo business casing (I grow revenue by 70% for two years, then level it off to 30% for the last two years; I grow staff up to 60 people max), you have something like this:
Those storage costs look insanely off. And from their press release, they claim to have actual data-centers (probably co-lo’d racks that are, at best, caged for compliance reasons, far from “having data centers”), which sounds like building your own, which might actually be as cheap, or slightly higher.
Who knows. Cloud storage is insanely cheap, so maybe that figure isn’t so bonkers. Of course, you need networking transfer chargers, etc. So, double, even quadruple the cost if you care too: still “nothing,” relative to the other numbers.
With this kind of Sunday morning, armchair analysis, there’s no end of flaws. Like I should have found the comparable costs, growth, the TAM, and staffing for Box, BackBlaze, etc., and even made sure I actually understand Spanning’s business model, but: ¯_(ツ)_/¯
Over years, that’s a pretty small gap to close to be profitably, and there’s a lot of things to play with in the spreadsheet (can we fire most all the sales and marketing people and go pure channel, hiring up a biz dev team of 2-3 people to get 5 or so key channel partners?).
It’s probably even easier to bundle up the company for sale to another large company after a few years. Someone like Microsoft or Salesforce might even want them to add that functionality to their own products, or any company that’s concerned about filling in it’s “enterprise SaaS” strategy gaps.
I’ve always like Spanning (RIP Sync). I hope it works out well!