Performance regression when restoring database with big metadata [CORE2634] #3042
Submitted by: @AlexPeshkoff
Relate to CORE3079
Restore of metadata only backup for database with about 100 000 records in RDB$FIELDS runs approximately 20-25 times longer on 2.X compared with 1.5.
According to Dmitry, the reason is new type of deferred work, added in 2.0, which is posted after any insert of the record into RDB$FIELDS table. Time is wasted when workign with DFW is stored in linear linked list.
The text was updated successfully, but these errors were encountered: