How to ingest 1B rows/s in ClickHouse | Hacker News Reader

How to ingest 1B rows/s in ClickHouse

tinybird.co

32 points_peregrine_10 comments

Minor quibble but: probably get another 5-15% by upgrading from oldoldstable Debian 11/Bullseye with kernel 5.10 to something modern like Debian 13/Trixie.

Edit: oh these are containers, right; the difference will be far less since the kernel is determined by the host. Still, I find it sad to see old software versions like this used.

> probably get another 5-15% by upgrading from oldoldstable Debian 11/Bullseye with kernel 5.10 to something modern like Debian 13/Trixie

For the on-prem/homelab crowd: I upgraded my desktop from Debian 12 to 13 yesterday only to discover my Mellanox ConnectX-3 NIC wasn't supported by the Trixie kernel. From 10GbE to 0GbE in one reboot :/

Have ordered a ConnectX-4 so that should get me going again, of course 100GbE is severe overkill for my needs but since I already have a 100GbE switch I figured why not...

Thanks.

I re-used all scripts and didn't even notice the linux version.

nice one

I don't know about ClickHouse specifically but these days it isn't that hard to ingest >10M rows/sec per server, including parsing, processing, indexing, and storage. You can just throw hardware at it.

yeah I mean that's basically what Javi talks about in the post... if you can throw hardware at it you can scale it (ingestion scales linearly with shards)

but the post has some interesting thoughts on how you do the high-scale ingestion while also handling background merge processes, reads, etc.

From yesterday https://news.ycombinator.com/item?id=44933452

definitely interesting and related

Is it just me or is there a noticeable uptick in Clickhouse related posts that are border-line marketing?

ClickHouse is just awesome and underrated. I tried to use TimeScale DB for 1-minute stock prices, and it was super slow and inefficient. On ClickHouse, the same data works, and it is blazingly fast.

Reply on news.ycombinator.com