Matthew O'Connor wrote:
>
> Could we have a counter in shared memory that keeps a count on the
> number seq scanners currently working on a table? While count = 1, we
> report all blocks regardless if it's from disk or from buffer. Once
> count > 1 we only report buffer reads. Would this have locking problems?
I don't know, but it would require some extra care to make sure that the
counter is decremented when a for example a transaction is aborted, or a
backend is killed.
> BTW, it seems to me that this is all based on the assumption that
> followers will have no problem keeping up with the pack leader. Suppose
> my process does a lot of other processing and can't keep up with the
> pack despite the fact that it's getting all it's data from the buffer.
> Now we have effectively have two different seq scans going on. Does my
> process need to recognize that it's not keeping up and not report it's
> blocks?
That's what I was wondering about all these schemes as well. What we
could do, is that instead of doing a sequential scan, each backend keeps
a bitmap of pages it has processed during the scan, and read the pages
in the order they're available in cache. If a backend misses a page in
the synchronized scan, for example, it could issue a random I/O after
reading all the other pages to fetch it separately, instead of starting
another seq scan at a different location and "derailing the train". I
don't know what the exact algorithm used to make decisions on when and
how to fetch each page would be, but the bitmaps would be in backend
private memory. And presumably it could be used with bitmap heap scans
as well.
-- Heikki Linnakangas EnterpriseDB http://www.enterprisedb.com