How big is the store on the drive? If it can’t fit everything in memory, it will have to evict something before it can write new data, which is typically appended to the ends of files. This eviction can become a bottleneck in situations with very high write loads, such as a benchmark that tries to push the database as fast as it can go. If you are hitting this bottleneck, then 2.3 might be good news for you, assuming your storage device is fast enough. No storage device is as fast as RAM, though, so allocating more page cache than you strictly need will still be faster.
-- Chris Vest System Engineer, Neo Technology [ skype: mr.chrisvest, twitter: chvest ] > On 08 Jun 2015, at 17:46, Frank Celler <[email protected]> wrote: > > thanks a lot for your improvements. I've updated the tests as follows: (1) > replaced the index by a constraint, (2) reduced the page cache memory to > 2.5GB, (3) updated to Java 8, (4) added a JIT warmup by executing 2500 > shortest paths before the test. This has improved the read results. However, > the write test got much worse. The culprit seems to (2). With 20GB page cache > memory the writes are almost twice as fast. -- You received this message because you are subscribed to the Google Groups "Neo4j" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. For more options, visit https://groups.google.com/d/optout.
