Oracle FAQ Your Portal to the Oracle Knowledge Grid
HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US
 

Home -> Community -> Usenet -> c.d.o.server -> Re: Row locking and serializability

Re: Row locking and serializability

From: David Cressey <dcressey_at_valinet.com>
Date: Sat, 12 Jun 1999 07:21:50 -0400
Message-ID: <_xr83.7715$G01.107914@news3.giganews.com>


Right you are. I certainly never intended the implication that Oracle and Rdb
function the same way. If anyone was likely to infer that from my remarks, your explicit statment to the contrary should help set them straight.

With regard to Oracle and snapshot transactions, I'm not following you. What the equivalent in Oracle of an Rdb snapshot transaction? I understand Rdb snapshots, but I don't see a similar construct in Oracle.

With regard to the substance of the paper you mentioned, I'm not going to ask for a summary here. I'll go off and read the paper itself, as time permits.

I agree with your statement that it's a complex subject. I think a lot of people underestimate the complexity.

Regards,

    Dave Cressey

Hal Berenson wrote in message ...
>Oracle and Rdb do not function the same way. In fact, Rdb did NOT allow
>updates (while Oracle does) inside a snapshot transaction precisely because
>such a scheme does not fully support the notion of serializability. There
>has been a large amount of debate about this ever since the ANSI SQL
>committee incorrectly described Serializability as part of the SQL92
>standard. The error crept into the standard as a result of trying to
>restate the requirements for serializability to not require pessimistic
>locking. This is a complex topic that was the subject of a paper at the
>1995 SIGMOD. Pat O'Neil and I initiated writing this paper, and very
>significantly the co-authors included Jim Gray (who authored the original
>paper describing serializability and isolation levels), Phil Bernstein (who
>has written more on this topic than any other author, and particularly
about
>optimistic techniques such as snapshots), and Jim Melton (who wrote the
>description of isolation levels that is in SQL92).
>
>You can find the paper (A Critique of ANSI SQL Isolation Levels) in the
>proceedings of SIGMOD '95 or at
>http://research.microsoft.com/scripts/pubDB/pubsasp.asp?RecordID=5
>
>--
>
Received on Sat Jun 12 1999 - 06:21:50 CDT

Original text of this message

HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US