3 ms·
My favorite was a DB project where the DB would accept DELETEs faster than the SSD could write them to disk. The eventual solution was to implement a proportion
by WJW 4y ago
My favorite was a DB project where the DB would accept DELETEs faster than the SSD could write them to disk. The eventual solution was to implement a proportional controller that looked at the "still to be persisted to disk" backlog and kept it at ~66% of the value at which the DB would stop accepting queries by selectively inserting sleeps into the DELETE-submitting process. We could have fixed it at some known-good but low value, but implementing a controller enabled us to speed up during the night (when there was low user traffic) and still not stress the db too much when a lot of users were online.
- twawaaay 4y agoYeah, I especially like to use CT for reducing configuration parameters or to replace meaningless configuration parameters (requests per second, batch sizes) with more meaningful ones (acceptable latency, acceptable failure rate).