google app engine - Improving database record retrieval throughput with appengine -
using appengine python , hrd retrieving records sequentially (via indexed field incrementing integer timestamp) 15,000 records returned in 30-45 seconds. (batching , limiting used.) did experiment doing queries on 2 instances in parallel still achieved same overall throughput.
is there way improve overall number without changing code? i'm hoping can pay more , better database throughput. (you can pay more bigger frontends didn't affect database throughput.)
we changing our code store multiple underlying data items in 1 database record, there short term workaround.
edit: these log records being downloaded system. fix in future , know how so, i'd rather work on more important things first.
try splitting records on different entity groups. might force them go different physical servers. read entity groups in parallel multiple threads or instances. using cache mght not work large tables.
Comments
Post a Comment