https://bugs.koha-community.org/bugzilla3/show_bug.cgi?id=17196 Olli-Antti Kivilahti <olli-antti.kivilahti@jns.fi> changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |olli-antti.kivilahti@jns.fi --- Comment #134 from Olli-Antti Kivilahti <olli-antti.kivilahti@jns.fi> --- I don't want to beat a dead horse, but can't resist:
There are several goals to do it: - Performance As Paul Poulain wrote, a simple query like SELECT publicationyear, count(publicationyear) FROM biblioitems GROUP BY publicationyear; takes more than 10min on a DB with more than 1M bibliographic records but only 3sec (!) on the same DB without the biblioitems.marcxml field
All our test environments: MariaDB [koha]> SELECT COUNT(*) FROM biblioitems; +----------+ | COUNT(*) | +----------+ | 1424537 | +----------+ 1 row in set (0.34 sec) On Ubuntu14.04 MariaDB using a normal 10k HDD with Intel Xeon 3GHz: innodb_buffer_pool_size = 128MB MariaDB [koha]> SELECT publicationyear, count(publicationyear) FROM biblioitems GROUP BY publicationyear; +-----------------+------------------------+ | publicationyear | count(publicationyear) | +-----------------+------------------------+ | NULL | 0 | +-----------------+------------------------+ 1 row in set (39.64 sec) The search is arguably slow. I wonder how the performance inspiration server (10 min search time) is configured? On Ubuntu14.04 MariaDB using RAID-1 SSD with Intel Xeon 3GHz 32GB RAM innodb_buffer_pool_size = 24G (Percona autoconfigurer) MariaDB [koha]> SELECT publicationyear, count(publicationyear) FROM biblioitems GROUP BY publicationyear; +-----------------+------------------------+ | publicationyear | count(publicationyear) | +-----------------+------------------------+ | NULL | 0 | +-----------------+------------------------+ 1 row in set (5.56 sec) RAM and SSDs are really cheap nowadays. See also Bug 18265 -- You are receiving this mail because: You are watching all bug changes.