Hacker News new | ask | show | jobs
by dominicl 1912 days ago
A definitive reference would be great, for me the "tried in the past" quote has always been a pointer to the Java VM. Which has great benefits of it's own but always had the problem of global GC induced freezes. Working with Cassandra instances this has always been a problem since 2013 till today. It just seems to be a very hard problem to solve when object references flow everywhere. But it's really unpleasant when your whole DB instance with all it's parallel requests locks up for GC.

[0] https://stackoverflow.com/questions/21992943/persistent-gc-i...
2 comments
secondcoming 1912 days ago
Try out Scylla instead. We moved to it from Cassandra and have had no stability issues at all. Our use case is several 100k requests per second with occasional batch writes. Request latency reduced, as did the number of nodes we required.
link
alexhutcheson 1912 days ago
DataStax had good results testing the new Shenandoah garbage collector with Cassandra 4.0: https://www.datastax.com/blog/apache-cassandra-benchmarking-...
link