chewbranca commented on issue #4650: URL: https://github.com/apache/couchdb/issues/4650#issuecomment-1599197142
@nickva nicely done demonstrating the lock collision overhead for the histograms. That said, it seems like the underlying histogram implementation is orthogonal to the issue of concurrent updates to the same histogram inducing locks. Your example for a histogram creates a single global histogram that would be concurrently updated just like we do with Folsom. So how do the alternatives to Folsom address the issue of concurrent lock collisions? For bucketing based histograms, given the same accuracy and range, they should be additive in theory. Were you hoping to achieve that with the histograms you linked? It's possible we could use a counter based ets backend for creating bucketing histograms, but counters are designed to be incremented and decremented, whereas with the sampling based histogram we use with Folsom will directly and go set the histogram reservoir values based on the current sample. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
