Coté

Coté

Waiting for the close of open - how long can the 2000s spirit of open source and open APIs last?

Midjourney: Medieval serfs defending a castle from demons in the style of Hieronymus Bosch, from venusinfurs.

The changing nature of “open”

In our tech world, the idea of “open” has changed a lot in the past few years. Instead of it meaning “open to everyone,” the classic notion of “open source,” now it more means “open to everyone except our competitors.”

Making money with open source is difficult

Running a high growth business on open source is difficult; you’re giving up on the easiest, most obvious thing to get paid for: the software itself. You can do an open core model where you are selling closed source software that uses/wrap around/ships with/improves/etc. the open source software. You can do the Red Hat thing where you sell support for your compiled open source project (and, I think, some closed-ish source “stuff” around it as well - this has become very confusing). But there’s not too much else. You have to come up with some excuse to charge people money for the free software. Or, the open source software can be part of an overall strategy that drives other revenue: the old razors and razor blades thing.

Once public cloud became mainstream, another model emerged: get paid to run and optimize the open source software. Run your Linux servers in the cloud, run your Postgres and Big Data stuff in the cloud, run your Spring Cloud stuff in the cloud, etc. This model works great for open source: the seller and the buyer get all of the benefits of open source and the seller can actually make money.

Attack of the channel!

The problem is that running a public cloud is expensive. Paying for all that infrastructure is expensive, and paying for the people who keep it up and running is expensive. So if each vendor wants to do it, they’re kind of stuck. They need to enter a new kind of business, being a managed service provider…a public cloud. Many have figured out this out as evidenced by the many SaaSes that are out there. I think most of them just use one of the public clouds and run their stuff on-top of it.

However, you then get into the problem of those public clouds wanting your business. Because you have open source, they could do exactly what you’re doing: sell operating the open source project. One of the first cases of this was Amazon running Elastic search. And there’ve been more, and there will likely be more. If there’s enough demand for running an open source project, the public cloud providers see an easy business model. And compared to their resources, the public cloud companies likely are better positioned.

To prevent this, many open source projects have tinkered with their licensing to basically say “this is totally open, except if you’re Amazon, Azure, or Google Cloud. But totally open and free for other people to use.” There’s all sorts of ways of enforcing this: it’s just fine-print shit, so who cares to catalog it?

Recently, Red Hat did this move as well, except it doesn’t really seem to be about public cloud providers. It was more focused on the classic problem of distributed open source projects, here, Linux. There were several “cloners” who would take the Red Hat Enterprise Linux source code, build it, and then sell it as a cheaper RHEL. Red Hat found a loop-hole (though, you only call it a loop hold if you don’t like it, otherwise, you just call “the license terms”) that would allow them to prevent their competitors from easily getting the source code. A close reading of the GPL license they use says that you only have to give the source code to customers, not everyone. So, Red Hat can decide that its competitors can not be customers, and, therefore, can not get the source code. Well, you can imagine the likes of Oracle, SUSE, and the others quickly figured out some tricks to get around that…sort of mooting the whole move by Red Hat?

Midjourney: Medieval serfs defending a castle from demons in the style of Hieronymus Bosch, from venusinfurs.

Closing Up Open Data and APIs

There’s another type of “open” that’s long been morphing. In the 2000s, the people building the web were very interested in open formats: XML, JSON…things that were basically plain text that you could read. They were also interested in fully open and free to use APIs for everything. Famously, Instagram used this openness to help build its social network. You’d create an Instagram account and you could use the Twitter API to suck in all your friends there. Over thousands and thousands of people, this means Instagram can start to duplicate all of the social networks inside Twitter, for “free.” Twitter no longer allows this, and few other social networks do as well.

And, most recently, Twitter has been just shutting down free API access.

In the post-Twitter world, there’s a lot of talking about ActivityPub (the open protocol used by Mastodon) and whatever alternate thing Bluesky uses. While ActivityPub works for Mastodon - and, like, is Mastodon - so far all the promises of open APIs in the mega post-Twitter Twitters has been just that: talk. Perhaps they’ll come out! When they do, we’ll need to see just how open and flexible they are. Could you write a complete clone of the Threads app with full functionality?

And, of course, Reddit also messed with its free APIs locking out alternate clients.

AI

Most recently, Facebook opened its LLM framework, but if I heard right, only for people that have less than 700,000 active monthly users. This means people like IBM could take it and sell it to even their largest customers: JP Morgan Chase, for example, has “only” around 300,000 employees, which is, [checks math] much less than 700,000. (Well, OK, except for use by every solider in the US military, etc., and you’d have to slice up the US government “customer” into agencies and groups of agencies.) But, tp the public cloud companies and Facebook, it’s much more closed than open.

And when I use ChatGPT, the URL includes the phrase “OpenAI.” But…that’s not the type of “open” I knew in the 2000s. Here, I think “open” more means “has an API that you can pay to use.”

Waiting for the close of open

My sense, then, is that the nature of “open” is changing. There is much resisting from us old guard people, but you can start to see the Planck Principle playing out, perhaps not in a morbid way but more in “how much energy do I have left to give a shit” way:

A new scientific truth does not triumph by convincing its opponents and making them see the light, but rather because its opponents eventually die and a new generation grows up that is familiar with it

Here, there’s not a “scientific truth,” per se, but more a principle of how the web wants to be. Rather, how the people building the web want it to be.

It feels like those people increasingly aren’t really interested in open.

Here’s some long running experiments to test that hunch:

  • Will Red Hat revenue from RHEL increase? If so, the users of RHEL care a lot less about “open.”

  • What will ActivityPub look like in Threads, Tumblr, and other mega sites? Will it all naturally fit together with Mastodon, or be, like, weird? If it’s weird, both the people building the web and the users will care less about “open.”

  • Will there be more cloud providers who have no commercial relationship with open source projects they run? Same.

We’ll see!

On-premises, air-gap Kubernetes Management

We have a great Kubernetes management tool, Tanzu Mission Control. But it’s only been available as a SaaS up to now. In enterprise sales, you always - always - eventually need an on-premises, off the Internet version. Militaries and spies use this, but there’s plenty of people who are heavily on-premises (based on four years of IDC estimates, I’d say that at least 50% of enterprise workloads are on-premises).

Anyhow! The most recent version of Tanzu Mission Control can run on-premises now. Air-gap and all that shit! Here’s an article about it, as well. Check out this interview I did with Corey Dinkens on the release. If you don’t like video, you can listen to the podcast.

Subscribe to this new .NET newsletter

My co-worker Layla started a new .Net newsletter, “Hooked on .NET.”

Subscribe if that’s your thing!

Who’s freaked out about AI, and who’s chill with it?

From Nicole Greene at Gartner: “38% of respondents to that same Gartner Consumer Community survey stated that they are very comfortable or somewhat comfortable, with an additional 27% taking a neutral position on the use of genAI in Marketing. That means 65% of consumers are mostly ok with marketers using genAI, and there are a lot of low risk ways to bring genAI to your marketing team right now.”

Wastebook

  • If you listen to the opening of “Breezin'” at half speed there’s some new magical thing that happens. Somehow it did this by accident today, and I thought I’d put on some kind of super high quality setting that let you hear the studio for real. It was a slow waking up of the song, like morning coming up.

  • “I mean, even if the Pope uses it, isn’t the purchase a bit expensive?”

  • “A gentle breeze blows, Birds singing in harmony, Nature’s symphony. DevOps buzz abounds, But failures linger around, Words won’t fix the sound.” Here.

  • Also: “continuously devops microserverless. with software and humans.”

  • ‘The lover of life makes the world his family,’ he wrote. ‘He is an “I”, insatiable in his appetite for the “not-I”.’ Here.

Relative to your interests

VMware Cross-Cloud Research and Insights | MENA

Things I had ChatGPT summarize for me: Identifying Key Industry Analysts; IBM Acquires Apptio for Hybrid Multi-Cloud FinOps; How Five Companies Built to Outcompete; Product and Platform Shift: Five Actions to Get the Transformation Right; Characterizing People as Non-Linear, 1st Order Components in Software Development; IncrementalOps: A New Approach to IT Operations; If Your Innovation Effort Isn't Working, Look at Who's on the Team; The GigaOm Pivot - Rebuilding the Analyst Business for the Digital Enterprise; What’s Wrong With the “What’s Wrong With Men” Discourse; China notes, July '23: on technological momentum; The New Media Goliaths; VMware’s Alex Barbato on How Agencies Could Enable DevSecOps Teams to Advance IT Modernization; VMware Tanzu Mission Control Self-Managed Announcement; Kubernetes: Innovation Enabler Or Implementation Detail?; Infosecurity Europe 2023, Forrester’s Thoughts; Is DevOps Tool Complexity Slowing Down Developer Velocity?; Dell Technologies Announces Intent to Acquire Moogsoft.

Upcoming

Talks I’ll be giving, places I’ll be, things I’ll be doing, etc.

July 19th Improving FinTech with cloud native think, speaking. July 19th Stop Tech Debt and Start Using Faster, More Secure Paths to Production. Sep 6th O’Reilly Infrastructure & Ops Superstream: Kubernetes, online, speaking. Sep 6th to 7th DevOpsDays Des Moines, speaking. Sep 13th, stackconf, Berlin. Sep 14th to 15th SREday, London, speaking Sep 18th to 19th SHIFT in Zadar, speaking. Oct 3rd Enterprise DevOps Techron, Utrecht, speaking.

Logoff

See y’all next time!

@cote@hachyderm.io, @cote@cote.io, @cote, https://proven.lol/a60da7, @cote@social.lol