Re: Serious article on comparison between MS SQL Server 2005 Yukon and Oracle 10g

From: DA Morgan <damorgan_at_x.washington.edu>
Date: Sun, 28 Nov 2004 13:21:15 -0800
Message-ID: <1101676782.826499@yasure>

Niall Litchfield wrote:

> "DA Morgan" <damorgan_at_x.washington.edu> wrote in message
> news:1101188533.245859_at_yasure...
>

>>A "transitory" latch is still serialization. That is substantially
>>different than a cached sequence.

>
>
> How? Oracle uses latching for sequences as well, and for the same reason. I
> guess one *might* be able to guarantee uniqueness by generating GUIDS based
> on session,timestamp and maybe some client info, but you aint going to get a
> 'nice' artificial incrementing key out of it.

Agreed. The question is really one of whether the point is to obtain an ordering such as 1,2,3,4 or whether guaranteed uniqueness is sufficient 1,3,2,4. GUIDs do the later but not the former. Sequences do the former but only if one forces serialization.

-- 
Daniel A. Morgan
University of Washington
damorgan_at_x.washington.edu
(replace 'x' with 'u' to respond)

Received on Sun Nov 28 2004 - 15:21:15 CST