… lot of time on sorts, more than on linux, so the configuration setting that was changed was the sort_buffer_size, we tuned it down from 2M to 1M, which somehow gave a big… hunderd of hours testing settings alone). The result in the graphs are of this testing, not only the result of the ‘one parameter…
Post: Star Schema Bechmark: InfoBright, InfiniDB and LucidDB
January 7, 2010 By 48 Comments
… did not expect I need big swap partition. Loading So load …- if put lineorder data in sorted order, loading time can decrease significantly…thread with results sometime better, sometime worse than InfoBright. Results: QueryInfoBrightInfiniDB LucidDB…realistic dataset and compare bigger set of …
Comment: Quick comparison of MyISAM, Infobright, and MonetDB
September 30, 2009 By 30 Comments
…mind what version you were using. For that sort of data set, even on the smallest Amazon VM, more… MyISAM, though. On the other hand, if that result came from enterprise edition eval version using INSERT … yet, a named pipe would be preferable for big imports. The difference query2 and query3 times is…
Comment: Sample datasets for benchmarking and testing
February 4, 2011 By 18 Comments
A big problem with these data sets are that they are small, trivial cases, which limits the … workload generators with parameter sets that exactly match the distribution, dynamics, and skew of real-world data sets. The results are not interesting per se but they allow you to characterize runtime behaviors under all sorts of assumptions…
Post: The two even more fundamental performance metrics
May 5, 2011 By 6 Comments
… throw away part of the data set to clean it up for modeling. … such as database servers. As a result, completion times shouldn’t be correlated either… other, but the completions will be sort of an echo of the arrivals, … actually dropped off sharply, leaving a big notch in the green curve, followed …
Post: Getting around optimizer limitations with an IN() list
January 9, 2010 By 9 Comments
… understand why these sorts of geographic search …4032 | Using where | +—-+————-+————-+——-+—————+———–+———+——+——+————-+ 1 row in set (0.00 sec) EXPLAIN … with a very big IN list. The … included in our results): SELECT * FROM coordinates …
Post: Watch out for Marketing benchmarks
August 23, 2006 By 7 Comments
… percent slower (especially due to handling multiple character sets). Of course I would not say it is … many types of subqueries, cases when hash or sort merge join is needed and in many other … to your application. The other big secret about benchmarks is – their results way to often have nothing to …
Post: The perils of InnoDB with Debian and startup scripts
January 28, 2009 By 13 Comments
…My chosen strategy for this was to set innodb_flush_method=O_DIRECT. We …user queries require it or not. On big servers I’ve seen it literally … 6344 | user | | NULL | 6345 | user | Updating | update … | 6346 | user | Sorting result | insert … | 6351 | user | | NULL | 6355 | user | | NULL | 6356 …
Post: MySQL Installation and upgrade scripts.
May 21, 2007 By 3 Comments
…perfect and I remember number of big issues when some releases could … by specifying utf8 as default character set together not to mention infamous …a lot of data type and sort order related changes, this is why… start after upgrade This is simply result of upgrade. Obviously with symlink removed…
