[Tickets #9342] Re: using an index on horde_cache table improve performance

[bugs at horde.org]

DO NOT REPLY TO THIS MESSAGE. THIS EMAIL ADDRESS IS NOT MONITORED.

Ticket URL: http://bugs.horde.org/ticket/9342
------------------------------------------------------------------------------
  Ticket             | 9342
  Updated By         | Michael Slusarz <slusarz at horde.org>
  Summary            | using an index on horde_cache table improve performance
  Queue              | Horde Framework Packages
  Version            | Git master
  Type               | Enhancement
-State              | Assigned
+State              | Feedback
  Priority           | 1. Low
  Milestone          |
  Patch              |
  Owners             | Horde Developers
------------------------------------------------------------------------------


Michael Slusarz <slusarz at horde.org> (2012-11-07 21:51) wrote:

In H5, horde_cache defines cache_id as an autoincrement key.  This  
(should?) automatically marks this field as a primary key.  (Testing  
on postgresql, attempting to alter the table and add cache_id causes a  
"multiple primary keys for table not allowed" error).

And I don't see how adding a unique index on cache_id and timestamp  
helps us.  We don't need to enforce the combo of those two fields to  
be unique.  And indexing on cache_id should be sufficient - SQL server  
should quickly look up cache_id, and then can quickly compare the  
timestamp field.

In other words, I think our cache table is already sufficiently optimized.