Observability Engineering

Well written. Helped accelerate my career. Recommending it to my colleagues.

Observability Engineering digs into what Observability actually _means_ (very very roughly "if you can answer novel questions to previously unknown problems without adding new telemetry of some kind, you have an observable system). It talks about the fundamental concepts, how it all works, how these systems are not technical, they are sociotechnical systems. And it talks about how observability leads naturally to a common language between devs and biz folks.That "sociotechnical" bit gets glossed over by lots of folks. If you're in a large company, and you're trying to convince management that they need to pony up some hard cash to build or buy observability? That's a social problem, not a technical one. Observability Engineering explicitly talks about this kind of stuff. It talks about the value add of observability in ways that will appeal to engineers, to product owners, as well as to budget holders.The book touches briefly on how you might go about building your own stack, albeit at a very high level.Lots of folks have complained about "it reads like an advert for Honeycomb" and... maybe they missed the parts where the authors very loudly proclaim where they work, and how that affects the book, and the parts where they talk Jaeger and Prometheus and Icinga and and and. Oh, and how they stress the _vendor neutral_ OTel standards they're pushing for.One of the little gems I discovered in the book is in the chapter "Forecasting to Create a Predictive Burn Alert". There's a bunch of stuff on forecasts of your error budget over differing time periods, and I realised that the same technique applies to ANY time series... and I happen to be working on some stuff that uses lots of time series.I heartily recommend this book to anyone interested in building systems that work for their customers.

This book makes the case for moving to use Observability rather than Metrics to manage large complex systems. It provides the theoretical underpinning of OpenTelemetry and the start of the practices needed to manage a complex system.In the past not having this level of sophistication I have been forced to simulate it by comparing relative counts of certain logs.It is written by the authors of one of the tools in the space (Honeycomb) but it is not Honeycomb specific.You would need to have written a tool of this type to understand the problem at the level that this book goes into.

ottima lettura per chi si confronta con applicazione fortemente integrate e distribuite

This book is a must-read for anyone looking to move away from merely correlating squiggly lines on a dashboard to find out if something is broken.It introduces a lot of ideas that companies which are now at a considerably scale are looking to do themselves. Structured events, sampling methods, issues with cardinality, SLO based alerting, are all fundamental things which need to be well explained in order for an organisation to do observability well; All of which are covered well in the book.