Gokulakannan Somsundaram wrote:
> Hi,
> The Current architecture of Updates in PostgreSQL is
> 1) Make a select query out of update. It involves a READ lock/BUFFER_SHARE
> 2) Get the tupleid
> 3) Goto the buffer containing the tupleid, make a BUFFER_EXCLUSIVE lock on
> it
> 4) update it
> 5) Repeat the above process for subsequent rows
>
> I propose to change this row-by-row approach, when it is a full table
> update. I plan to send a extra flag(which will be set for Full table
> Deletes/Updates). this would make the access method directly acquire the
> exclusive lock and update the existing record.
>
> For Deletes this is simple. But for updates, the projection tuple has to be
> made before re-inserting it. So there will be a list of Heap tuples stored
> in memory for each page getting updated. these tuples will be inserted after
> the deletion part of update is done. This is just a rough design. I may get
> involved in a detail design once i get a nod from the mailing list
> community.
I doubt the locking overhead is that significant. Have you done any
profiling to show that it's worth it?
-- Heikki Linnakangas EnterpriseDB http://www.enterprisedb.com