Hi,
Looked into one topic more, see comment below.
jey Razack wrote:
> Listed below are the required features to manage
> >1TB VLDBs:
>
> - Global index on the primary table (across
> partitions)
> - Local indexes on each partitions.
> - Select fld1,... from
> <PrimaryTable>.<PartitionName> WHERE
> ....
> - sqldump : dump entire table or selected
> partitions
> - import entire table or selected partitions.
> - Move partition from one tablespace to
> another/new.
> - Merge,split , rename,truncate and drop
> partitions. Also set status to active or
> inactive.
> - Option to load any partition to cache (heap
> storage engine may not help)
> - Able to create view on a set of selected
> partitions.
> - Optimization hints for Index usage and other
> select / join operations.
> - Don't allow to drop all the partitions. Atleast
> the table must have one active partition.
> - Row counts are cached for each partitions and
> for the entire table so that
> the select count(*) from
> <table>.<partition> will display the
> count faster.
Whether row counts are cached for each partition depends on the underlying handler.
(This is controlled by the table flag HA_NOT_EXACT_COUNT in the handler interface)
for MyISAM this is the case but for most others it is not. The partitioned handler will
make use of the cached row count if available on the partitions below.
Rgrds Mikael
>
> Regards..
> Jey Razack
Mikael Ronstrom
Senior Software Architect, MySQL AB
My blog:
http://mikaelronstrom.blogspot.com