Re: Db2, Oracle, SQL Server

snip

Serge Rielau wrote:
> Noons wrote:
> > hpuxrac apparently said,on my timestamp of 19/02/2005 1:25 AM:
> >
> >> The process occurs whether you know about it or not and is used
and
> >> described in oracle rac documentation.
> >
> >
> > And any node can be the master for the duration
> > of the recovery. It's only been like that since V6.2
> > in 1989 when I first installed and used OPS. What
> > else can I say...
> >

RAC of course is a huge improvement over OPS which didn't exactly meet a lot of customer expectations after the marketing hype of OPS went south.

> Maybe you should:
> http://download-west.oracle.com/docs/cd/B10501_01/rac.920/a96596.pdf
> page 146 in the PDF file. Figure 7-1
> Maybe it's outdated or plain wrong. Can't find it in O10g.
> The metalink notes may help here (don't have access :-)
> That btw may explain the difference of opinion.
>
> >> It depends on what kinds of accesses are being attempted against
any
> >> data affected for as long as instance recovery is occurring.
> >
> >
> > As opposed to NO ACCESS at all in other databases.
> Yeah that would be shocking.
> Good enough
> > for me.
> >

RAC works fine for what it does. TAF has some serious issues.

How long you get impacted during an instance failure, if you get impacted at all ... it all depends.

Just trying to keep the discussion honest.

http://www-306.ibm.com/software/data/highlights/rac.pdf

The IBM url above appears to be a marketing push for DB2 and as such cannot be trusted very far. But most oracle professionals would agree there are some thoughts in that document that are pertinent.

It's been a long time since I was hands on with DB2. It was a fine product then and is I am sure vastly improved, just as oracle has been improved. Received on Fri Feb 18 2005 - 16:40:47 CST