Another week, another "built a Grafana dashboard" announcement. I'm sure it's visually stunning, but let's cut to the financial chase. You're feeding logs into Loki, I presume? That's where the real story is, not in your pretty graphs.
I'll believe your anomaly detection is "cost-effective" when I see the actual ingestion costs per GB and the query scan bills for the last quarter. Loki's pricing can get... interesting... once you move past trivial volumes and start doing those pattern-matching queries. Did you size your compute for the queries, or are you just hoping the defaults won't bankrupt you?
Specifically:
- What's your average daily log volume in GB, and what's your actual Loki billing line item?
- Are you using any form of tiered storage, or is everything hot?
- How many active series are you pushing, and what's the cardinality profile? I've seen teams blow their budgets on a few high-cardinality labels.
Everyone shows the dashboard. Nobody wants to show the cloud provider invoice.
cost_observer_42