2016-04-25 Noel Grandin <[email protected]>: > 2016-04-25 Nicolas Barbier <[email protected]>: > >> Would you be interested in me posting screenshots of non-MVCC >> profiling results for 1.3.176 vs. 1.4.191? Or maybe 1.3.176 vs. >> nightly build? > > Sure, thanks, that would be interesting. Also try running EXPLAIN PLAN on > both for the relevant queries, just to make sure we aren't doing something > silly there.
I just redid the tests with 1.3.176 and 1.4.191, and I got identifical timings of about 20s. I also did the same tests when using server mode, and got about 43s with both versions. So I guess that before I must have accidentally enabled the server mode instead of embedded mode. Conclusions: (1) It seems to be that 1.4.191 is indeed extremely much slower than 1.3.176 at our kind of operations (bulk insertions in a PageStore database), but only when using MVCC. When MVCC is disabled on 1.4.191, the performance is the same. (2) The server mode (where the server runs on the same machine as the client) seems to be about half as fast for this kind of operations, in both versions 1.3.176 and 1.4.191. (3) The problem where we sometimes got a network protocol error (when using the server mode; as reported by Rob in the thread “H2 Server mode and blobs: Unknown operation: -46695848”) that exists in 1.3.176 seems to be fixed in 1.4.191. Thanks for your input, Nicolas -- A. Because it breaks the logical sequence of discussion. Q. Why is top posting bad? -- You received this message because you are subscribed to the Google Groups "H2 Database" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To post to this group, send email to [email protected]. Visit this group at https://groups.google.com/group/h2-database. For more options, visit https://groups.google.com/d/optout.
