Re: how to build a database from scratch

From: NENASHI, Tegiri <tnmail42_at_gmail.com>
Date: Fri, 8 Dec 2006 16:53:58 +0100 (CET)
Message-ID: <Xns98936F04E699asdgba_at_194.177.96.26>


"Cimode" <cimode_at_hotmail.com> wrote in news:1165586735.555464.27590_at_80g2000cwy.googlegroups.com:

>
> NENASHI, Tegiri a écrit :
>
>> Sampo Syreeni <decoy_at_iki.fi> wrote in
>> news:Pine.SOL.4.62.0612080913040.17498_at_kruuna.helsinki.fi:
>>
>> >
>> > 2PL does make life easy, but you can't use it unless you know
>> > precisely what data is going to be touched by a transaction when it
>> > begins.
>>
>> It is very strange what you say. Microsoft SQL Server utilizes 2PL
>> and I think DB2 also. But one does not "know precisely what data is
>> going to be touched". Then how it is possible that many of people
>> use these databases ?
>>
>> --
>> Tegi
>
> SQL and DB2 have been implementing ORACLE Read Consistency since the
> SQL2005 and UDB 8.2.

Is it that before 2005 SQLServer could not be used ? Or DB2 before 8.2 ? It was speculated that "you can't use it unless you know precisely what data is going to be touched".

The deadlocks. SQL Server 2005 with Snapshot, that I think is the same like Read consistency of Oracle, can still have deadlocks because it uses locks when one upates rows.

>It essentially insures that in any context, the
> last COMMITTED version of the data is ALWAYS available independently
> from the transactional context that affects the tables.
>
Received on Fri Dec 08 2006 - 16:53:58 CET

Original text of this message