Hacker News Reader

We (Livepeer) switched to self-hosted Loki recently for this and couldn't be happier. Completely S3-backed, cheapest option possible, adjusting your retention is as easy as setting an expiration rule on the bucket. Queries are as fast as the hardware you feed it 'cause it's just all your nodes doing a brute-force search.

Query language is a bit of a PITA compared to some of the other options, but you get used to it.

View on HN

Interesting - didn’t know it could by S3 backed. I shall have to investigate that!

You can get the whole grafana stack to be s3 backed now I think.

Cortex for Prometheus, and tempo also all have s3 storage capabilities iirc

You can also use Thanos (https://thanos.io/) for storing Prometheus metrics into S3. The design is pretty similar to Loki (which Thanos predates), separated into services so you can nicely horizontally scale your ingestion path separately from your read path depending on workloads.

Reply on news.ycombinator.com