On Fri, 23 Dec 2005, Tom Lane wrote:
> Qingqing Zhou <zhouqq(at)cs(dot)toronto(dot)edu> writes:
> > I revised the idea with MINIMAL XLOG (instead of WITHOUT XLOG) like the
> > below. I think in this way, we can always gaurantee its correctness and
> > can always improve it.
> I think the entire idea is a waste of time anyway. If we have the COPY
> case covered then we have done the only case that has any practical use
> AFAICS --- you don't use UPDATE or DELETE for bulk loads, and you don't
> use retail INSERTs either if you are striving for speed.
There are several posts on the list asking about NOLOGGING option
presented in Oracle. User may need it to do bulk updates against the
database. I don't think we plan to support it since it does not gaurantee
transaction semantics. But MINIMAL XLOG is something that we are afford to
do and not invasive change needed AFAICS.
In response to
pgsql-hackers by date
|Next:||From: Greg Stark||Date: 2005-12-24 03:41:42|
|Subject: Re: [Bizgres-general] WAL bypass for INSERT, UPDATE and|
|Previous:||From: Tom Lane||Date: 2005-12-24 03:00:39|
|Subject: Re: [Bizgres-general] WAL bypass for INSERT, UPDATE and |