The more kubernetes you use, the more frequently you release software
and other insights from the latest CNCF kubernetes survey, a new cfgmgmt company, on-premises kubernetes pricing strategies, and links.
The more kubernetes you use, the more frequently you release software
The CNCF has a new kubernetes survey out. We discussed it in more length in today’s Tanzu Talk News (it’s not published yet, but subscribe to it and you’ll get it tomorrow). Here’s some notes of mine:
I did not make the time to read the raw data. When I have in the past, it’s always been a treat.
Demographics: 2,063 took it. I generally consider this survey to be “advanced” users. For example, they put out calls on their listservs, Twitter accounts, etc. - so they’re recruiting from their community. But they’ve done some things to “normie” it: pulling in Dynatrace and Data Dog scanning, using an external firm to find respondents. I would still consider this survey to represent “the CNCF community” more so than “everyone.”
For example, Gartner estimated that in 2022 5% of all terrestrial workloads ran on kubernetes.
That disclaimer aside, findings in this survey from the CNCF community:
The way they word the following is confusing. “44% of respondents are already using containers for nearly all applications and business segments and another 35% say containers are used for at least a few production applications.” But then “30% of our respondents' organizations have adopted cloud native approaches across nearly all development and deployment activities.” // I think they’re saying 44% of respondents run most of their apps in production on kubernetes, 35% run at least 3–5 (but not 50% to 100%?) in production. And, more people are running their apps not in production, in dev and staging. As mentioned, I haven’t looked at the detailed data set so perhaps this is in there, but I’d want to just straight up ask “what percentage of your apps in production run on kubernetes.”
The more you use kubernetes, the closer to deploying at will (multiple times a day) you get: “organizations that widely use cloud native approaches, 76% are using containers for nearly all applications and business segments, and 48% release code at least daily. In comparison, only 20% of organizations with limited cloud native maturity use containers and 23% release applications daily.” // there’s a 9% rise in multiple times a day releases if you’re in the “nearly all” category.
The “operating system of the cloud” observation is interesting. We talked about it a lot in today’s Tanzu Talk recording (posted on Feb 8th, 2023, not the day of this newsletter episode - “we’re taping it now, but it airs next week” and all that). What this means is that people are just using kubernetes as the thing they run everything on, well, like an operating system. They’ll run their build pipelines in it, their IDP’s like Backstage, etc. This is a really interesting aspect of kubernetes and one that’s maybe not…expected? It’s not just a production “cloud,” but just a think you run stuff in. I don’t know how to articulate that well. A good way of putting it from the survey: “At Dynatrace, we use Kubernetes for any new software project, whether it’s build pipelines or our SaaS offerings. We see the same trend with our customers. Kubernetes effectively has emerged as the operating system for the cloud.”
“Auxiliary workloads” is what they call this stuff.
From the CNCF 2022 survey. 63% of kubernetes workloads are these auxiliary ones, with 37% being application workloads, in 2022. The report summary doesn’t specify if the app workloads are in production, but let’s assume that for squinting at this rainbow chart purposes, it doesn’t matter. Other than the observation itself, this shows some spend-thinking CIOs (or whoever) will need to do: you could easily end up paying more for development and staging than production if your vendor doesn’t distinguish between the two. I don’t think the public clouds care: it’s all kubernetes to them. But, private cloud vendors could negotiate that as part of the EULA (or whatever) pricing. You’d have to run on trust, largely, but vendors who only charged for production kubernetes workloads would have a pricing advantage over public cloud. The public cloud people need to account for IaaS and networking usage - they pay for that (as little as it maybe be!). But software vendors (yes, like us at VMware Tanzu) have nothing to do with that. So, if we just said, “we’re not going to charge for anything but production workloads,” we could, using the numbers from the auxiliary workloads rainbow, give people a big discount on their potential spend. Of course, the public cloud services with money to burn on long-term CAC-think could do the same and just eat the cost until they starved us out. STRATEGY! FUN!
There’s a huge rise in serverless usage, from 2020 to 2022: “service mesh going from 27% in 2020 to 47% in 2022, and serverless architecture/FaaS moving from 30% to 53%.” // Ever since Simon Wardley started declaring next year the year of serverless about five (six?) years ago, we’ve all been waiting for it. I floated a theory once that we’ve reached “peak serverless,” that we’ve figured out all the uses. But, it seems to be growing y/y/y. What will it be next year. Also, if throw in things like GitHub Actions (that’s serverless right?) the number goes up. Again, I’d want the question to be “how many of your production workloads use serverless?”
Check the rest, I love a good survey.
SPEAKING OF! Are you getting kubernetes setup in your organization? Check out the methodology we’ve used for a decade or more to help large organizations run their application platform, platform as a product. We have an overview documented in this PDF that will give you a good start. It’s the topic of the “platform” talk I give, so if you want to pull on that string, there’s plenty more out there. Download it it for free.
My Content
Two Guys Live in the Southern Hemisphere, with Craig Box, Software Defined Talk - This week Matt Ray is joined by Craig Box and they discuss living down under, Craig’s media future, and managing CNCF Sandbox projects developer relations.
State of Spring 2022 Survey, Kubernetes in the Wild, Tanzu Talk News podcast - In this episode, we check in on the Spring community by looking at the most recent State of Spring survey. While in survey land, we look at the Dynatrace Kubernetes in the Wild study: what are the top languages used in kubernetes, and how is on-premises use shaping up> Also, SpringOne was great, and now there’s so much more of it each week until the next SpringOne this fall. And, check out how smooth Ben is at picking up MC duties when Coté’s internet stops working.
cfgmgmtcamp 2023 trip report
Admittedly, not including my own talk, I only attended two talks (see below) at this year’s configuration management camp. And, I’ve had to skip the second day since I have more travel this week.
I saw Adam Jacob’s talk on his new company, System Initiative which seems like applying an object oriented approach to configuration management. It has a clever use of digital twins to basically have a simulation of production in your staying environment and then, cleverly, to sync production back to your staging model. This means that your CMDB/one source of truth “lies less,” as he put it.
The idea and architecture are great! As always it gets down to the “drivers”: whatever actually connected to those Cisco routers and databases and does the changes and sends those changes back to the staging model. I’m always a bit let down that people don’t spend more time on this stuff. Eventually you have to get down to the ugly sendmail configuration that Adam started with.
In the long term vision, you have the hardware and software people make these “drivers,” which I guess is what is natural order of things if you can get them onboards.
Anyhow, it seemed like a good idea. And the talk was great, you should watch it for a good story of fixing a server when you’re wasted.
What ends up happening, every year so far (or least the last two), is that I get delightfully stuck in the cafeteria talking for hours with someone. Last time is was my old pal John Willis, this time it was with Colin Humphreys, Winna Bridgewater, and a couple other folks. We talked about what their company is up to, and, eventually, Descartes.
My talk was OK. It’ll be better next time as I rev the new stuff some more. I spent too much time defining and talking about platforms and not the promise of the talk: what to actually do. That’s easy enough to cut down and, then, expand on the good parts. I enjoyed talking with people afterwards, as always.
Wastebook
Culture without tools it’s just talking.
“JS-industrial-complex” Here.
“Heroku has all the hallmarks of a good platform: it’s simple, it works and its way too expensive” Me, apparently.
Relevant to your interests
Resetting professional goals - “The traditional definition of SMART goals is Specific, Measurable, Achievable, Relevant, and Time-Bound. It’s a useful framework in itself, in the sense that at least it provides some structure and concreteness, but in its definitions it encourages you to diminish mission and values in your work. A goal that is Specific and Achievable is highly likely to just be iterative on what you’re already doing. In contrast, SMARTIE encourages you to dream. By replacing Specific with Strategic, we’re encouraged to think longer-term. By replacing Achievable with Ambitious, we’re given permission to stretch for what we really want. Swapping Relevant for Realistic reminds us to keep our feet on the ground, but in paired this with Achievable we’re being asked to find a productive middle ground between our dreams and reality. And then reminding us that our goals must be Inclusive and Equitable ensures that we consider our impact on others, and on our communities and ecosystems.”
What ChatGPT Can’t Tell Us About the Future of Healthcare - “So when there is (A) widespread agreement on something, and (B) that something is often written about, the mimic [AI] is likely to be correct. If those two conditions are not true … not so much.” And, quoted from Paul Graham, how to be, in the words of Tyler Cowen “more weird”: “Idea: Before writing an essay, have ChatGPT write one on the same topic to show you what would be the conventional thing to say, so you can avoid saying that.”
The Locksmith Paradox - Relevant to pricing enterprise software and infrastructure, and it also shows you why it’s hard to sell something as “easy” and “productive” unless the customer has first tried to DIY it themselves. // “He was tipped better when he was an apprentice and it took him longer to pick a lock, even though he would often break the lock! Now that it takes him only a moment, his customers complain that he is overcharging and they don’t tip him. What this reveals is that consumers don’t value goods and services solely by their utility, benefit from the service, but also a sense of fairness relating to how much effort was exerted.”
A Treatise Concerning the Properties and Effects of Coffee (1792) - “Soon he too ‘began to dance and frisk about as the goats had done.’”
Yes, Crypto is ALL a Scam - “a scam in the truest sense. Sell the illiterate hoi polloi on an imagined fear built around a complex narrative and then profit from that fear in a self-sustaining closed loop that generates its own demand”
Reduce-Kubernetes-spend-with-these-10-Kubecost-alternatives - List of Kubernetes cost monitoring and management things.
Two paths for AI journalism - ‘As for CNET, the Verge reported that its “AI system was always faster than human writers at generating stories, the company found, but editing its work took much longer than editing a real staffer’s copy.” In other words, if you want to create writing that is useful to humans, you need to invest in a significant amount of human work somewhere along the line, if not in the writing of the text, at least in the editing of it.’ And: ‘Apparently as a consequence, Buzzfeed’s stock nearly quadrupled in two days. Granted, we’re talking about a stock that was trading at 95 cents last Wednesday and now goes for around two bucks. But its sudden rise suggests, if nothing else, that “AI” and related terms are right where “web3” or “crypto” were a year or two ago in terms of “buzzwords that, if used strategically in announcements, memos, and public interviews, will pump up your stock price.”’
Logoff
I have another trip tomorrow, up to London. I have some thoughts on iA Presenter I’ll have to share once I find the time. See y’all next time!