Currently the over flow table of flow table is a different table and is
searched if lookup does not succed in main table. The over flow table is
searched completely starting from a hash till the entry is found. This
leads to performance drop if entry does not exists in the table as such.
As a fix, the flow table is adapted to generic hash table, which
maintains a linked list of entries from the last bucket entry of main
table. So the search always looks into only these entries rather the
complete table.
Reviewed: https:/ /review. opencontrail. org/14315 github. org/Juniper/ contrail- vrouter/ commit/ 9737c778e7d56f2 71d1f3480861b39 e8a193c91b
Committed: http://
Submitter: Zuul
Branch: R2.20
commit 9737c778e7d56f2 71d1f3480861b39 e8a193c91b
Author: Divakar <email address hidden>
Date: Thu Oct 8 16:05:48 2015 +0530
Adapting flow table to generic hash table
Currently the over flow table of flow table is a different table and is
searched if lookup does not succed in main table. The over flow table is
searched completely starting from a hash till the entry is found. This
leads to performance drop if entry does not exists in the table as such.
As a fix, the flow table is adapted to generic hash table, which
maintains a linked list of entries from the last bucket entry of main
table. So the search always looks into only these entries rather the
complete table.
Change-Id: Icc4bd90a194696 708b6ac847ee67a 4b1b4256c8b
closes-bug: #1497833