Published: March 5, 2020, 10 p.m.
Cot\xe9 probably messed up his math on the Thoma Bravo profit from Compuware. Maybe it's more like $5bn. But, obviously, he's just farting around with incomplete information. He apologies and will sit in the corner for awhile. For entertainment only!
\n\n
With the virus shutting down conferences and keeping people in the home office, we discuss the value of in-person conferences and how remote ones might could be better. Also, GKE\u2019s kubernetes cluster pricing (and Amazon\u2019s drop to match the price) gives us an anchoring point for pricing running a cluster. Coupled with the recent CNCF survey you could make an interesting stew. Finally, Cot\xe9 tries to run some numbers to figure out how much Thoma Bravo profited from taking Compuware private. (Also, he always mispronounces it as Thom-oh Bravo.)
\n\n
Hey! If you want to understand VMware\u2019s new strategy and portfolio around application development, tune into the March 10th webinar on the topic: register now!
\n\n
Mood Board:
\n\n
\n- The Hello Boss episode.
\n- You mean I gave my kids a lecture all for nothing?
\n- Shit the living room door.
\n- Espresso macchiato.
\n- Blue bonnet coffee.
\n- What am I missing out on?
\n- Sitting, trapped in your head.
\n- I\u2019m hoping we realize no one needs to be working.
\n- Why blockchain is important for corn
\n- Containers and the mainframes,
\n- Too unqualified to speculate?
\n- This post-conference era
\n- You can\u2019t do your dishes in the office
\n- Those poor developers: having to pay for things!
\n- $879/year per cluster helps someone keep their job.
\n- It\u2019s always Monte Carlo simulations with you
\n- So adult.
\n- I\u2019m trying Matt Ray.
\n- We IPO\u2019d because we were running out of money, what a time to be alive.
\n- This week\u2019s white guys talking about white guys.
\n
\n\nRelevant to your interests\n\n
\n- How Marc Benioff\u2019s Vision for Salesforce\u2019s Future Triggered Executive Shuffle
\n- VMware exceeds $10B in sales in FY 2020
\n- Cisco begins new round of layoffs
\n- How much money do SREs make?
\n- Coronavirus 2020 tech conference cancellations list \n\n
\n- BMC to Acquire Compuware\n\n
\n- Agile software development is dead. Deal with it.
\n- Coronavirus Updates: Epidemic Slows in China but Spreads Globally - Check out the Ali app angle
\n- Apple to pay up to $500 million to settle lawsuit over slow iPhones
\n- Google makes Hangouts Meet features free in the wake of coronavirus
\n- CNCF survey\n\n
\n- No lead-gen on the PDF! CLASSY.
\n- Demographics:\n\n
\n- \u201cSeptember and October 2019 and received 1,337 responses.\u201d
\n- 30% of respondents from orgs with 5,000+ employees.
\n- ??? \u201cThe top job functions were software architect (41%), DevOps manager (39%), and back-end developer (24%)\u201d
\n- Most respondents from \u201cSoftware,\u201d \u201cTechnology,\u201d and \u201cFinancial Services\u201d - all the bleeding edge.
\n
\n- Amazon is #1, Google probably #2.
\n- CI/CD (loosely applied) is at 40% to 50% - which is close Cot\xe9\u2019s ongoing estimates (and, considering that most of the respondents are from tech and banks, if we\u2019re cynical, probably less for the other industries).
\n- The jump in production is really quick, maybe (Page 5, \u201cUse of Containers since 2016\u201d? It took about 4 years for prod use to be broadly done (in Dec 17, prod reached 75% which matches test)\n
\n- Most figures like these (e.g., number of containers in production) would be a lot more interesting/useful if they were broken out by company size. E.g., larger companies probably use more containers in production, tech and banks probably have put containers in production earlier, also telcos - T-Mobile alone has 34,000 containers in production (probably even more by now).
\n- Similarly, how many clusters are in production would be interesting to see by organization size.\n\n
\n- Challenges are sort of interesting, as always. \n\n
\n- I don\u2019t like \u201cculture\u201d as a broad category. That usually just means \u201cpeople don\u2019t do what I think they should do [and instead have their own ideas of what\u2019s best].\u201d
\n- However: obviously \u201csecurity\u201d\u2026\u201dcomplexity\u201d is another broad category - and, boy, long-time SDT sponsors must love \u201cmonitoring\u201d as a money-pot to go after!
\n
\n- Side-note: so, \u201cservishmesh\u201d means a registry to look-up how to connect to other pods/components in your kubes (like, JNDI); getting the actual network connection to that other component; securing the network connection; load balancing (this term is getting way over-blown, I think?); and then doing the layer whatever networking to account for dynamically assigned IP addresses and stuff in kubernetes. Maybe, like microservices stuff like circuit breakers, or is that too far?
\n- Not that many people use their own serverless framework (10%), but 34% of those who do use knative.
\n- The \u201cwhy you use kubernetes\u201d chart (pg. 11) didn\u2019t force people to rank enough: pretty much everyone agrees that All The Value-Props are great.
\n- Helm wins for packaging.\n
\n- I don\u2019t know enough about auto-scaling to say much, but it looks like most people don\u2019t do auto-scaling unless it\u2019s for purely stateless apps, which makes sense. The drop-off after that (queues, batch-jobs, stateless, and DB) seems to indicate that auto-scaling other stuff is difficult, untrusted.
\n- \u201cnginx kept its lead this year as the top Kubernetes ingress provider (62%), followed again by HAProxy (22%)\u201d - F5 got a good control-point on the kubernetes market for $670 million, plus the entire rest of the nginx business.
\n- \u201c40% of respondents get their info from Twitter\u201d - humanity had a good run!
\n
\n
\n\nNon Sense\n\n
\n\n
Arrested DevOps Podcast:
\nSubscribe today by searching for \u201cArrested DevOps\u201d in you favorite podcast app or by visiting https://www.arresteddevops.com/.
\n\nConferences, et. al.\n\n
\n\nSDT news & hype\n\n
\n\nRecommendations\n\n
\n\n
Cover art from Marcingietorigie in wikicommons.