On 2/16/19 10:36 AM, Vladimir Sitnikov wrote:
> Benjamin> A related and helpful patch would be to capture the access log and
> Benjamin> provide anonymized traces.
>
> The traces can be captured via DTrace scripts, so no patch is required here.
>
Right. Or a BPF on reasonably new linux kernels.
> For instance:
> https://www.postgresql.org/message-id/CAB%3DJe-F_BhGfBu1sO1H7u_XMtvak%3DBQtuJFyv8cfjGBRp7Q_yA%40mail.gmail.com
> or
> https://www.postgresql.org/message-id/CAH2-WzmbUWKvCqjDycpCOSF%3D%3DPEswVf6WtVutgm9efohH0NfHA%40mail.gmail.com
>
> The missing bit is a database with more-or-less relevant workload.
>
I think it'd be sufficient (or at least reasonable first step) to get
traces from workloads regularly used for benchmarking (different flavors
of pgbench workload, YCSB, TPC-H/TPC-DS and perhaps something else).
A good algorithm has to perform well in those anyway, and applications
generally can be modeled as a mix of those simple workloads.
regards
--
Tomas Vondra http://www.2ndQuadrant.com
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services