so annoyed by all the "topic: observability!" followed by chatter about logs and monitoring. that is not going to help you with your unknown-unknowns, it's o11y-washing old shit to sell it to you.
i made two checklists. grab a marker, let's figure out what you're listening to --
i made two checklists. grab a marker, let's figure out what you're listening to --
Technical observability:
❑ high cardinality
❑ high dimensionality
❑ wide, rich events (aka spans)
❑ schemas are inferred and dynamic
❑ no friction to adding more context
❑ unique ids, propagated
❑ SLOs
❑ tracing
❑ column store or autoindexed
❑ very near realtime
❑ high cardinality
❑ high dimensionality
❑ wide, rich events (aka spans)
❑ schemas are inferred and dynamic
❑ no friction to adding more context
❑ unique ids, propagated
❑ SLOs
❑ tracing
❑ column store or autoindexed
❑ very near realtime
Marketing "observability":
❑ pillars (one for each product they're trying to sell you)
❑ metrics
❑ AIOps (sic)
❑ logs
❑ pricing based on hosts or seats
❑ dashboards
❑ magical autoinstrumentation
❑ tags
❑ single pane of glass
❑ ML, AI
❑ aggregates
❑ strings, searching
❑ pillars (one for each product they're trying to sell you)
❑ metrics
❑ AIOps (sic)
❑ logs
❑ pricing based on hosts or seats
❑ dashboards
❑ magical autoinstrumentation
❑ tags
❑ single pane of glass
❑ ML, AI
❑ aggregates
❑ strings, searching
ten years from now i'm just gonna be a crank waving my cane and grumbling about why you still haven't structured your logs 
