In this post we will walk through an investigation of latency spikes observed in the ScyllaDB 1.0.x version family, which are linked to the scylla-jmx service running alongside the ScyllaDB server on the same system. We will show how the scheduler can be tuned to reduce latency spikes. ScyllaDB 1.2, which is at the release candidate stage this week, will have all relevant fixes. Users of that version won’t have to do anything. ScyllaDB 1.2 is scheduled for release in about a week. It all started with a report from one of our users. They tested a read-only and very … Continue reading Reducing latency spikes by tuning the CPU scheduler