RE: 64 node Oracle RAC Cluster (The reality of...)

I believe you are discussing two different things:

1. Which instance will write out the next batch of dirty blocks for some file
2. Which instance's memory contains the particular read consistent vintage of a block needed by some client.

If I understand it correctly, you are each very close to a useful explanation of the order of operation of these two different aspects of multinode communications requirements in RAC.

In other news, Mladen's note about functional partitioning is of course the best way to design systems for RAC. Oracle's marketing point is not that it is not good to functionally partition, just that you don't HAVE TO re-write your applications.

Regards,

mwf

-----Original Message-----
From: oracle-l-bounce_at_freelists.org
[mailto:oracle-l-bounce_at_freelists.org]On Behalf Of K Gopalakrishnan Sent: Tuesday, June 21, 2005 9:53 AM
To: mcdonald.connor_at_gmail.com
Cc: oracle-l_at_freelists.org
Subject: Re: 64 node Oracle RAC Cluster (The reality of...)

Connor:

I am failing to connect your thoughts. If there are n number of nodes, the resources will be mastered at n number of nodes, thus reducing the contention for mastership (okay,, there is nothing called contention for mastership, I am just thinking theoritically) and this will help in getting required ownership transfers faster as one instance will have limited number of resources. In crude terms we can compare this with multiple freelists.

Btw, dynamic remastering does not work at block level. They work at file level and/or file level/object level.

On 6/21/05, Connor McDonald <mcdonald.connor_at_gmail.com> wrote:
> I haven't thought this through much, but assuming clients are
> connecting to instances at "random", then isn't the probability of a
> instance requring a block for which it is already the holder approx
> 1/n, where n is the number of the nodes. So as the node count rises,
> you'd get a reduced likelihood of "success" ?
>
> I suppose dynamic remastering is meant to assist with this as well.
>
> Connor
>
> On 6/21/05, K Gopalakrishnan <kaygopal_at_gmail.com> wrote:
> > Raj:
> >
> > There won't be technically any differnt from 3 node to 64 node as in
> > RAC a maximum of 3 parties involved in ANY resource management
> > (Master-holder-requester) . Be it is 3 nodes or 64 nodes or 128 nodes.
> >

--

Best Regards,
K Gopalakrishnan
Co-Author: Oracle Wait Interface, Oracle Press 2004
http://www.amazon.com/exec/obidos/tg/detail/-/007222729X/
--
http://www.freelists.org/webpage/oracle-l


--
http://www.freelists.org/webpage/oracle-l