Thread: Re: [HACKERS] GSoC 2017 Proposal
Hello, Mark!
On Mon, Apr 3, 2017 at 7:00 PM, Mark Rofail <markm.rofail@gmail.com> wrote:
------
Alexander Korotkov
Postgres Professional: http://www.postgrespro.com
The Russian Postgres Company
Kindly find my proposal attached to this email.
I'd like to ask what do you mean in research item number 3?
3. Making the full-table sequential scan GIN-indexable instead seems very reasonable since GIN is primarily used to search for element values (PK values) that appear within composite items (FK array).
a. Statistics have shown[7] that GIN indexing an array shows an increase in performance by ~2256% !
b. Thus the first step (as proposed ) would be to prove that “<@( is contained by)” can be used in this scope.
What exactly you're going to do in this item? Check that GIN index over array support given operator, do the performance benchmark or something else?
------
Alexander Korotkov
Postgres Professional: http://www.postgrespro.com
The Russian Postgres Company
Dear Mr Alexander, I was checking the archives today and to my shock, I did not find my reply to your previous question which was almost two weeks ago. I apologise for the inconvenience, I have however replied within an hour but apparently, it did not go through. Best Regards, Mark Moheb On Thu, Apr 6, 2017 at 4:18 PM, Mark Rofail <markm.rofail@gmail.com> wrote: > Hello Mr Alexander, > > From my understanding, the main issue occurs whenever any UPDATE or > DELETE statement is executed on the PK table, > this triggers a referential integrity check on the FK table. In the > previous patches, this was done by performing a sequential scan. > > To improve performance I propose that we index the FK column, and in > my point of view the most suitable index would be the GIN index since > it is targeted for composite items. > However, to move forward with this approach, we have to be sure that > the comparison semantics offered by GIN indexes satisfy our needs for > the referential integrity check. > > This approach was proposed by Tom Lane: > https://www.postgresql.org/message-id/28389.1351094795%40sss.pgh.pa.us > > I believe this can be accomplished by better understanding the GIN > index implementation in postgreSQL, including its operators. > > This is the best to the knowledge I gained during the application > period. I would like to investigate it further and would be delighted > to hear your input regarding the matter, > > Best Regards, > Mark Moheb