Trying to migrate to 0.9.11 found changes in update_partition_stats (described here).
Adding mandatory ANALYZE on every update_partition_stats leads to long execution time.
This affects both the time of operations with items and the time of bulk insertion of records when migrating data from a file.
As an example: I have DB instance with 16Gb RAM and collection with ~30000 items. Request to add new item, update or delete takes 3 minutes of time, specifically, for the ANALYZE operation. More powerful machine with 32 RAM, performs ANALYSIS in 5 seconds on same amount of records, but this is not what you expect from production server with 30000 rows table.
Perhaps this feature should be made optional.
Trying to migrate to 0.9.11 found changes in update_partition_stats (described here).
Adding mandatory ANALYZE on every update_partition_stats leads to long execution time.
This affects both the time of operations with items and the time of bulk insertion of records when migrating data from a file.
As an example: I have DB instance with 16Gb RAM and collection with ~30000 items. Request to add new item, update or delete takes 3 minutes of time, specifically, for the ANALYZE operation. More powerful machine with 32 RAM, performs ANALYSIS in 5 seconds on same amount of records, but this is not what you expect from production server with 30000 rows table.
Perhaps this feature should be made optional.