Reducer #5: Observability in the Matrix & delivery pipelines
Engineering
This week I did some catch up with the O11ycast; here are the highlights:
Ep. #37, Synchronization Layers and Logs With Garry Shutler of Cronofy @ O11ycast
Can you see the woman in the red dress?
Garry gives a great example of how challenging it is to distribute knowledge about complex systems. It's common for early team members to be the only ones to resolve most of the incidents.
Garry:
It's like being in the movie The Matrix and I see all the green symbols and the woman in the red dress, but everyone else can't.
Charity:
It's like when you are the person who built the system from the ground up, you know, you get this sixth sense, you can't even explain it sometimes just.
Software that you have written and not your deployed ages, like Fine milk
Software delivery is like digestion. It's like one of my favorite quotations in the last couple of months was: software that you have written and didn't deploy ages like Fine milk. It does because you have all this context in your head, you know why you built it, you know what you did? You know what you tried, you know the whole path, the trade-offs, but it decays really rapidly. Like you've got it all in your head for what minutes? Maybe an hour or two. When you move to the next project, you've evacuated all that past knowledge, and your head is in the next set of problems.
You need to ship one engineers changes at a time, which is how you get software ownership over changes. It is reasonable to ask an engineer if you merge to main in a few minutes, you go and you look at it. If there are 10 different engineers changes in the deployment, who is in charge of looking it up?
Ep. #38, Infrastructure Changes With Andy Davies of Reaktor @ O11ycast
If your build times are slow, it doesn't mean you need Microservices
Fast build times are the result of many small optimizations. Try parallelizing tests, instrument the pipeline, visualize bottlenecks and work on them before thinking about Microservices. Don't assume you know; instrument first, use data to guide your decisions.
Photo by Lazarescu Alexandra on Unsplash
Writing
Twitter thread with a framework to generate interesting ideas for content.
Every online writer struggles to: • Generate ideas consistently • Turn ideas into interesting content • Get people to actually read their work Until they learn these 3 game-changing frameworks & create their own Endless Idea Generator. I wish I knew this sooner 👇🏼 (View Tweet)
Datadog content strategy
How @datadoghq got so good at content-driven growth: they hired full-time engineers to write blog posts. (View Tweet)