Charity Majors @mipsytipsy CTO @honeycombio, ex-Parse, Facebook, Linden Lab; cowrote Database Reliability Engineering; loves whiskey, rainbows. I test in production and so do you. 🌈🖤 May. 24, 2019 1 min read

Rearchitecting storage assumptions, reconceptualixing UX and migrating legacy data so your service can deliver observability for users is *extremely* hard and expensive.

Fortunately, changing your marketing copy to *say* you do observability is easy and cheap! 🤑🤥🤯

Y'all never even get to hear most of our users' most glowing quotes about @honeycombio.

Because it's usually paired with an exasperated kick at their old tools. And we don't post quotes that refer negatively to other tools in the space.

Because we aren't really going head to head with any of them; we occupy space at the nexus of monitoring, logging, tracing, and APM.

(I think these categories have a short shelf life, and will be absorbed into a single observability category in 3-5 years.)

We aren't competing with other tools: we are competing with the old way of building software: at arms length, using rusty fishing poles instead of fingers, with a blindfold cinched over your eyes and all the ensuing pain buffered by an ops team.

It can be better. 💖🌈🐝

But in order for a better world to emerge, it is *critical* that we not get twisted on the definition of observability.

It will set us back as an industry if the vendors succeed in muddling a cornerstone technical term and rendering it toothless. So ask them these questions:

* can I query across the original raw event data?
* only structured data?
* can events be arbitrarily wide?
* can you add/remove dimensions on the fly?
* are all dimensions high cardinality? Can I break down by a high cardinality dimension?
* are there no indexes?


* are data points organized as context for an event (not disconnected and tagged like metrics)?
* is the data being emitted from within your code, not gathered by an observer?
* is time a first class citizen, and can temporally related events be reconstructed?


* do multiple requests over an interval get aggregated, represented as one number?
* are percentile buckets written out at write time or calculated afresh at read time?
* are your graphs interactive, and can you drill down from high level to individual req?


You can follow @mipsytipsy.


Tip: mention @threader_app on a Twitter thread with the keyword “compile” to get a link to it.

Enjoy Threader? Sign up.

Since you’re here...

... we’re asking visitors like you to make a contribution to support this independent project. In these uncertain times, access to information is vital. Threader gets 1,000,000+ visits a month and our iOS Twitter client was featured as an App of the Day by Apple. Your financial support will help two developers to keep working on this app. Everyone’s contribution, big or small, is so valuable. Support Threader by becoming premium or by donating on PayPal. Thank you.