Re: effective_io_concurrency's steampunk spindle maths

On Sat, Mar 7, 2020 at 11:54 PM Evgeniy Shishkin <itparanoia(at)gmail(dot)com> wrote:
> > On Mar 7, 2020, at 00:33, Thomas Munro <thomas(dot)munro(at)gmail(dot)com> wrote:
> > That is indeed what led me to start thinking about what a good new
> > name would be.
>
> MySQL has a term io_capacity.
> https://dev.mysql.com/doc/refman/8.0/en/innodb-configuring-io-capacity.html
> > The innodb_io_capacity variable defines the overall I/O capacity available to InnoDB. It should be set to approximately the number of I/O operations that the system can perform per second (IOPS). When innodb_io_capacity is set, InnoDB estimates the I/O bandwidth available for background tasks based on the set value.
> >
>
> Perhaps we can have maintenance_io_capacity as well.

That sounds like total I/O capacity for your system that will be
shared out for various tasks, which would definitely be nice to have,
but here we're talking about a simpler per-operation settings. What
we have is a bit like work_mem (a memory limit used for each
individual hash, sort, tuplestore, ...), compared to a hypothetical
whole-system memory budget (which would definitely also be nice to
have).