3 ms·
I processed ~3m messages a day on Sqlite using a pair of ca 2006 era Xeon on spinning rust (for redundancy; each could easily handle the load by itself). The qu
by vidarh 4y ago
I processed ~3m messages a day on Sqlite using a pair of ca 2006 era Xeon on spinning rust (for redundancy; each could easily handle the load by itself). The queue processor was written in Ruby and ran on the (very slow) 1.8.x series (even then it used about 10% of a single core.
On modern hardware you should be able to trivially handle more than that.
- randito 4y agoDoes this mean you are processing messages on only one machine, since it's Sqlite? Depending on what you are processing, that could take longer than the queue/dequeue.
- vidarh 4y agoThe queue was managed by a server written in Ruby that spoke Stomp [1], so while for the biggest queue most of the processing did in fact happen on the same machine, that was just because it happened to fit. [1] https://stomp.github.io/ https://stomp.github.io/