Steven Lembark <lembark@wrkhors.com> writes:
> On Fri, 27 Apr 2018 19:38:15 +0300
> Vitaliy Garnashevich <vgarnashevich@gmail.com> wrote:
>
>> We're going to try using "SELECT 1 FROM table FOR KEY SHARE" for each of
>> the tables, which are referenced by results, before running the big
>> query. That should be up to a million of rows in total. It will probably
>> not cover the case when a record is INSERT'ed and then DELETE'd after
>> the calculation has begun, but such cases should be even more rare than
>> the DELETE's we're currently facing.
>
> Thing about using a couple of Materialized Views for the worst
> part of it.
+1 re: materialised views - I have found them to be extremely useful for
situations where you want a snapshot of data and need to present it in a
way which is easier to process, especially when the underlying data is
changing faster than your reporting process can generate the report.
--
Tim Cross