> This pattern is not only theoretical, the talk cites public examples from Instacart, which has described PostgreSQL on NVMe for a latency-sensitive search workload
article basically teaches that computers with local storage has better latency. Wisdom of back to basics.
Not that it teaches a bad thing or in a bad way, just me expected from Clickhouse guys less obvious things
Consumer grade SSDs 100-500 iops, datacenter SSDs in thousands (like ~ 5 year old samsung model on Hetzher AFAIR had ~ 3000 iops in that test). Cheap VPSes - easily in 5-30 iops of this sort
more often then not, fsync/fdatasync is overlooked by programmers
> This pattern is not only theoretical, the talk cites public examples from Instacart, which has described PostgreSQL on NVMe for a latency-sensitive search workload
article basically teaches that computers with local storage has better latency. Wisdom of back to basics.
Not that it teaches a bad thing or in a bad way, just me expected from Clickhouse guys less obvious things
reply