,
7 tweets,
3 min read
Read on Twitter
HOLY ROYAL SHITBALLS. To all of you who have been asking (for YEARS) if there are any alternatives to honeycomb for observability: yes, finally YES!
@will_sargent has hacked together the most ingenious little solution using only logs and feature flags:
tersesystems.com/blog/2019/07/2…
@will_sargent has hacked together the most ingenious little solution using only logs and feature flags:
tersesystems.com/blog/2019/07/2…
I am plumb smuckered with joy. This is so fucking clever. This makes a great poor man's o11y option, or live debugger in prod, and it's just a great tool to keep in your pocket.
He calls it "diagnostic logging", or "logging with an audience" (which is....ouch lol)
He calls it "diagnostic logging", or "logging with an audience" (which is....ouch lol)
"If business logic and data is blood, then observability is lymph. Imagine circulatory system and the lymphatic system, running in parallel to each other. Lymph is hidden and never seen, but it monitors and protects at all times." 🔥🔥🔥 
He goes on to crisply lay out the nature of distributed systems problems, and why metrics, logs, and traces DO NOT solve the problem (and ARE NOT magical pillars that transmogrify into observability).
For most mere mortals this would be more than enough, BUT WAIT THERE'S MORE
For most mere mortals this would be more than enough, BUT WAIT THERE'S MORE
Because @will_sargent still has energy to school you in twenty odd years of "why logs are terrible" and "99 ways logs will hurt you (and everyone you love)".
And then show you how to wizard shit up with feature flags.
And then show you how to wizard shit up with feature flags.
In conclusion, no, you CANNOT get observability by purchasing datadog, or newrelic, or splunk, or all three. No matter how much money you shovel their way.
Their marketing is /actively/ misleading you.
Their marketing is /actively/ misleading you.
But you *can* now get barebones observability with feature flags and log libraries, thanks to @will_sargent. 👏🌈🌷
