michael wilde
@michaelwilde.bsky.social
X beckons you back!
March 12, 2025 at 4:21 AM
X beckons you back!
It’s mostly possible if you do tail based sampling (traces are the most popular thing to use for event based SLOs in HNY because spans are the only type of event that requires a duration). In fact, the economics of o11y SaaS pretty much require you to sample.
December 2, 2024 at 11:57 AM
It’s mostly possible if you do tail based sampling (traces are the most popular thing to use for event based SLOs in HNY because spans are the only type of event that requires a duration). In fact, the economics of o11y SaaS pretty much require you to sample.
Yes. That’s actually what Honeycomb’s SLO product does. It’s SLI is based on actual events versus observed time aggregations that are taught in the Google SRE handbook. It’s a different level of detail, which is effective for many folks.
December 2, 2024 at 11:42 AM
Yes. That’s actually what Honeycomb’s SLO product does. It’s SLI is based on actual events versus observed time aggregations that are taught in the Google SRE handbook. It’s a different level of detail, which is effective for many folks.
Why not for Kafka? Actually interested. Thx
December 2, 2024 at 2:37 AM
Why not for Kafka? Actually interested. Thx
97% correct. The other 3% is infra metrics are a thing as platform eng/infra peeps are the formula 1 engineers of resources. Nothing wrong with looking at charts of non-transactional metrics. Can’t derive everything from events. (yet)
December 2, 2024 at 2:10 AM
97% correct. The other 3% is infra metrics are a thing as platform eng/infra peeps are the formula 1 engineers of resources. Nothing wrong with looking at charts of non-transactional metrics. Can’t derive everything from events. (yet)
Terminal colors are lit
December 2, 2024 at 1:59 AM
Terminal colors are lit