Turns difficult cloud native concepts into easy to use infrastructure. It’s shameless, blameless, inclusive, for all ages and fun to listen to. You’re guaranteed to learn some new concepts or ideas on every show.
How Ahrefs Saved US$400M in 3 Years by NOT Going to the Cloud
AWS lays off another 9,000
Kubernetes & cloud native operations survey
Listener feedback from the podcast and Youtube
How Ahrefs Saved US$400M in 3 Years by NOT Going to the Cloud 37Signal wasn’t the first company to complain about the very high cost of cloud and they won’t be the last. Case in point this week is Singaporean company Ahrefs, coming with a real banger and stating that they “saved” $400M in expenses by NOT going to the cloud.
AWS lays off another 9,000 The recent mass layoffs in the industry make for a better scenario to hire cloud talent. Behemoth Amazon, employing about 1.6M people in 2022 will add over 9000 people to that talent pool shortly. CEO Andy Jassy said in a memo to staff on Monday that as part “of the second phase of our operating plan (“OP2”) this past week, I’m writing to share that we intend to eliminate about 9,000 more positions in the next few weeks—mostly in AWS, PXT, Advertising, and Twitch.”
Kubernetes & cloud native operations survey Admit it, you’ve tried to buy a Raspberry Pi4 now for a long time and have given up. Playing the lottery might be a winning strategy here, so why don’t you try entering the Kubernetes & cloud native operations survey.
I’ve talked about it in episode 3. There is an interest in figuring out how to avoid huge cloud bills and if there’s a way to run workloads that make sense locally. There are some clear low-hanging fruits, like storage that causes a lot of charges for outbound traffic, or AI workloads that are CPU intensive and require expensive instances. So when I came across this article on “Are companies shifting away from public clouds?” I was intrigued. What do the good folks at StackOverflow (by the way, they have an excellent podcast, you should listen to it if you don’t already do, go find it in the show notes or head over to https://stackoverflow.blog/podcast/ ) think about this?
Cloud Native Security
I work in the security space at IBM and mostly deal with what I call “traditional security” on the infrastructure level. But Kubernetes security is an entirely different beast altogether. Why is it so different? It’s because the whole stack needs to be looked at, not just the application layer 7. Yes, mostly K8 communicates via APIs on layer 7 and pretty much all the magic happens there, but there’s still the network layer, there’s still the unsolved question of trustworthy build pipelines using public images and many more issues. So where do you even start?
How long does it take to spin up a K8 cluster on a cloud provider from scratch?
So let me be honest here and start out by saying creating a K8 cluster in our lab will take far, far longer than even the slowest cloud provider takes to create a cluster. But how quickly can AWS, Azure or Google actually get a K8 cluster ready for you?
Listener feedback from the podcast and Youtube
I absolutely love listener feedback and it’s really nice to hear from you! Thank you to Hanna in Oregon who writes “I was going through apple podcasts yesterday looking for cloud and openshift content when I discovered the open cloud infrastructure podcast which has valuable content. I’m listening to an episode about running databases on containers. Congrats!”
Thank you Hanna!
And here’s a question from my YouTube channel: “
Oscar Llerena • 19 hours ago
Thanks a lot for your video. I would like to ask you for some directions. I was inspired by your setup to start buying a small scale similar one to implement my home lab … but a friend working on devops striked me with a question of why not do this on cloud … I replied that on Cloud I will have to pay monthly tens of dollars for the computing. RAM, and diskspace that I am targeting for my projects .. and he replied that I will have to continuously invest on hardware too, as my computational needs will increment in the future …. what do you think. Thanks in advance.”
Here’s what I think, Oscar. Personal labs aren’t used for benchmarking or for performance testing. At least that’s not what I do in my lab. The majority of lab work consists of figuring stuff out, like networking, storage setup and general compute tasks, like installing operating systems, virtualization or application stacks. None of that is really time sensitive. And in fact my hardware is really old already. The G7’s came out in 2010 and my machines are roughly 10 years old. My G8s are a little younger, but essentially obsolete machines for production. But they still work, they still run just fine, they still provide me with a super-cheap opportunity to play and get stuff working. I don’t upgrade them because there’s no more need to do that. I don’t think you’ll need to continuously upgrade your lab. A one-time buy should last you many years. Thanks for your question, keep them coming!
