On Tue, 2018-06-05 at 05:57 -0700, Andres Freund wrote:
> But I think my proposal to continue use a hashtable for the already
> known groups, and sorting for additional groups would largely address
> that largely, right? We couldn't deal with groups becoming too
> large,
> but easily with the number of groups becoming too large.
The eviction problem is the same whether we partition or sort: which
groups do we keep in memory, and which ones do we send to disk?
You are essentially suggesting that we do what my patch already does
w.r.t eviction: don't evict; the groups that appear first stay in
memory for the duration, later groups may be forced to disk.
Regards,
Jeff Davis