"We completely and utterly fucked it up by defining observability to mean “gigachad-scale JSON logs parser with a fancy search engine.” Really? Really? That’s the “we solve Real Serious Business Problems™” strategy we went with?"
@drmorr@hazelweakly Lots of useful ideas here, but I still prefer the control theory definitions. Can you see what’s going on is observability. Can you figure out what to do about it is modelability but people don’t seem call it that. Can you change the system is controllability. That’s a feedback loop. The whole thing is resilience. Modelability is where most of the human aspects come in.
@adrianco@drmorr I think the reason people don't use modelability is because it sounds backwards in a sense. Or it's not entirely obvious what makes sense there?
I think really something I could've articulated better is that "distributed tracing != Observability" and somehow the control theory definition got very tightly bundled with distributed tracing.
Q: doesn't resilience also want adaptive capacity? Or is that implied by controllability somehow?
@adrianco@drmorr that said, it does sound like the (data + human bits + action) collection of concerns is the thing that is kinda missing and that a lot of people do agree it exists. Whether or not you call all of that observability or just part of it is a different story :)
@hazelweakly@drmorr Model-ability: can you make sense of the observations. That depends on training, experience, up to date run books. Whether someone managed to code an automated model that works well enough most of the time. (E.g. CPU is high so increase autoscaler limit). Tracing is a really good observability upgrade as long as you can digest the results. You have resilience if the system is “in control” and working inside its capacity.
@adrianco@drmorr@hazelweakly I prefer the term understandability to modelability just because it’s more familiar and hence simpler. Understandability has always been the missing link in observability (monitoring) systems. It does no good to collect tons of information if humans don’t understand what it means and what control actions it should imply.
Add comment