From: | Amit Kapila <amit(dot)kapila16(at)gmail(dot)com> |
---|---|
To: | Mithun Cy <mithun(dot)cy(at)enterprisedb(dot)com> |
Cc: | pgsql-hackers <pgsql-hackers(at)postgresql(dot)org> |
Subject: | Re: [POC] A better way to expand hash indexes. |
Date: | 2017-02-20 16:21:28 |
Message-ID: | CAA4eK1K6eAEcoTTJh1yDWVD65v7AaavnjHOqEczwq0U1qTX27A@mail.gmail.com |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-hackers |
On Fri, Feb 17, 2017 at 7:21 PM, Mithun Cy <mithun(dot)cy(at)enterprisedb(dot)com> wrote:
>
> To implement such an incremental addition of bucket blocks I have to
> increase the size of array hashm_spares in meta page by four times.
> Also, mapping methods which map a total number of buckets to a
> split-point position of hashm_spares array, need to be changed. These
> changes create backward compatibility issues.
>
How will high and lowmask calculations work in this new strategy?
Till now they always work on doubling strategy and I don't see you
have changed anything related to that code. Check below places.
_hash_metapinit()
{
..
/*
* We initialize the index with N buckets, 0 .. N-1, occupying physical
* blocks 1 to N. The first freespace bitmap page is in block N+1. Since
* N is a power of 2, we can set the masks this way:
*/
metap->hashm_maxbucket = metap->hashm_lowmask = num_buckets - 1;
metap->hashm_highmask = (num_buckets << 1) - 1;
..
}
_hash_expandtable()
{
..
if (new_bucket > metap->hashm_highmask)
{
/* Starting a new doubling */
metap->hashm_lowmask = metap->hashm_highmask;
metap->hashm_highmask = new_bucket | metap->hashm_lowmask;
}
..
}
Till now, we have worked hard for not changing the page format in a
backward incompatible way, so it will be better if we could find some
way of doing this without changing the meta page format in a backward
incompatible way. Have you considered to store some information in
shared memory based on which we can decide how much percentage of
buckets are allocated in current table half? I think we might be able
to construct this information after crash recovery as well.
--
With Regards,
Amit Kapila.
EnterpriseDB: http://www.enterprisedb.com
From | Date | Subject | |
---|---|---|---|
Next Message | Peter Moser | 2017-02-20 16:38:10 | Re: [PROPOSAL] Temporal query processing with range types |
Previous Message | Christoph Berg | 2017-02-20 16:15:56 | Re: powerpc(32) point/polygon regression failures on Debian Jessie |