Royce Ausburn a écrit :
> All of the queries on this table are reporting on a single collection, so ideally a collection's data would all be
storedin the same part of the disk... or at least clumped together. This can be achieved using "cluster", however as
faras I know there's no automated, non-cronesque means of clustering and having the table become unusable during the
clusteris not ideal.
>
>
>
If the lock level used by CLUSTER is a problem for you, you could
consider pg_reorg contrib. AFAIK, it does similar work as CLUSTER but
allowing a concurrent read and write activity on the table.
Regards. Philippe.