Re: Proposal: global index

On Aug 21, 2017 07:47, "Simon Riggs" <simon(at)2ndquadrant(dot)com> wrote:

On 18 August 2017 at 15:40, Alvaro Herrera <alvherre(at)2ndquadrant(dot)com> wrote:
> Ildar Musin wrote:
>
>> While we've been developing pg_pathman extension one of the most frequent
>> questions we got from our users was about global index support. We cannot
>> provide it within an extension. And I couldn't find any recent discussion
>> about someone implementing it. So I'm thinking about giving it a shot and
>> start working on a patch for postgres.
>>
>> One possible solution is to create an extended version of item pointer
which
>> would store relation oid along with block number and position:
>
> I've been playing with the index code in order to allow indirect tuples,
> which are stored in a format different from IndexTupleData.
>
> I've been adding an "InMemoryIndexTuple" (maybe there's a better name)
> which internally has pointers to both IndexTupleData and
> IndirectIndexTupleData, which makes it easier to pass around the index
> tuple in either format.

> It's very easy to add an OID to that struct,
> which then allows to include the OID in either an indirect index tuple
> or a regular one.

If there is a unique index then there is no need for that. Additional
data to the index makes it even bigger and even less useful, so we
need to count that as a further disadvantage of global indexes.

I have a very clear statement from a customer recently that "We will
never use global indexes", based upon their absolute uselessness in
Oracle.

It is worth noting that the only use case I can see where global indexes
fill a functionality gap are with unique indexes which allow you to enforce
uniqueness across an inheritance tree where the uniqueness is orthogonal to
any partition key.

I could find large numbers of uses for that. That could also allow
referential integrity to check against a root table rather than force
partition explosion.Will

Otherwise the following mostly works:

Create table (like foo including all) inherits (foo);

So the gap this addresses is very real even if it is narrow.

> Then, wherever we're using IndexTupleData in the index AM code, we would
> replace it with InMemoryIndexTuple. This should satisfy both your use
> case and mine.

Global indexes are a subset of indirect indexes use case but luckily
not the only use.

--
Simon Riggs http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services

--
Sent via pgsql-hackers mailing list (pgsql-hackers(at)postgresql(dot)org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers