We’ve been working on our new Hibernate second level cache implementation at Terracotta for the upcoming 3.1 release. I was doing some perf testing today just to kick the tires a bit and took this screenshot of a small part of the cache tooling (all part of the open source project btw):
In this screen shot you’re seeing:
- Cache Hit Ratio graph – showing you the aggregate second-level cache hit ratio, here looking pretty good.
- Cache Hit/Miss Rate – showing you both the hit and miss rates (the watermark is the latest value, constantly updated)
- DB SQL Execution Rate – showing you the rate queries are being executed against the database. This is not the lame query execution rate you can get from the Hibernate statistics that only shows some queries – rather, this is showing ALL queries against the database.
- Cache Put Rate – hopefully, the obvious rate of puts into the cache. Here we’re doing a fiar amoutn of loading.
- Region table – at the bottom, a per-region table showing you a further breakdown. For each cache region, you can see the hit ratio, hits, misses, and puts. Also, the in-memory count (only valid if you change the Select View control in upper left to a particular cluster node) and total count of entries in the cache. And the last two columns show you the average latency in the case of a hit against the clustered cache and a miss which must load from the database.
Good stuff, and a big help in tuning.