Skip to content

Comment on How RocksDB Worksparent

Comments

I didn't catch the part where "Parent wants storage at the source node.".

So if the goal is to eventually have the timeseries data merged back to a timeseries DB, and latency isn't too much of a concern then wouldn't batch writing to Kafka (Kinesis, etc) be tolerable?

AboutSource Built by g1lg1l

Hackerly is an independent reader for Hacker News, built on the public HN API. Not affiliated with Y Combinator.