RE: Q. To RAC or go vertical

Yeah, I thought about making the statement a little more vague to leave myself plenty of wiggle room. But, what the heck, go ahead and make it, and see what happens.

Granted, there might be extreme circumstances in which RAC might be the only way to bring sufficient numbers of CPUs to bear. But, in general, while I'm sure Oracle would much rather you pay $60K per CPU than $40K, and in order to get people to do that, they would like everyone to believe that RAC is that amazing application with a million-and-one uses, I can't think of any situation where two boxes with, for example, 8 CPUs each would match the performance of one box with 16 CPUs. assuming all other things are equal, and we don't have some pathological situation like the old Sparc 2000 that could not support the number of CPUs it could hold. To put it more generally, I don't think scaling out is as good as scaling up when it comes to best database performance.

If the economics of the situation were such that a Redundant Array of Inexpensive Computers (RAIC??) were the goal -- i.e. multiple cheap computers as opposed to one very pricey computer -- and if your application writers had the intellectual kahones and considerable programming experience to write the application to efficiently use such an arrangement, then RAC would make sense. In the context of the original post, I got the impression that the company has fairly deep pockets, so there is no intent to use RAIC as a bargain basement equivalent of a big box (as far as I could tell). There was no indication that the application(s) will be tailored to RAC, so I don't think scaling out will be as effective as scaling up.

If fault tolerance is a requirement for the installation, or finances require some kind of RAIC installation, or you have maxxed out your E15K, then RAC might be appropriate. But I don't think its use as just an alternative means of bringing in more CPUs is correct.

And one of the other guys here just said Oracle now has the top TMP(c or d) benchmark using HP Itanium systems. I haven't checked it out yet, so that's just hearsay. I guess that means Microsoft will now be required to put together a bigger cluster to beat it.

> -----Original Message-----
>
> While fault tolerance is certainly one of the features of RAC,
> it isn't correct to say that it is not also for scalability.
>
> Buy a bigger box? That works fine until you're in the biggest
> box you can get, then what? I realize that it's a small market
> segment that requires that kind of hardware, but it still exists.
>
> Sun has been testing a cluster of 15k servers with RAC, ostensibly
> for scalability. Some nodes are populated with 78 CPU's and 288
> Gig of RAM. ( yes, that is correct ).
>
> Jared
>
>

-- 
Please see the official ORACLE-L FAQ: http://www.orafaq.net
-- 
Author: Stephen Lee
  INET: Stephen.Lee_at_DTAG.Com

Fat City Network Services    -- 858-538-5051 http://www.fatcity.com
San Diego, California        -- Mailing list and web hosting services
---------------------------------------------------------------------
To REMOVE yourself from this mailing list, send an E-Mail message
to: ListGuru_at_fatcity.com (note EXACT spelling of 'ListGuru') and in
the message BODY, include a line containing: UNSUB ORACLE-L
(or the name of mailing list you want to be removed from).  You may
also send the HELP command for other information (like subscribing).