Re: Why doesn't Oracle care about Linux as IBM does?

On Fri, 17 Aug 2001 03:22:47 -0500, leebert <leebertarian_at_yahoo.com> wrote:

>
>No doubt the transparency of OPS & DB2 MVS is nicer, but the management
>overheads of OPS and mainframes are notorious. Pick your trade-offs.
             

I dunno how you got the "management overheads of OPS and mainframes" in the same sentence, but let me assure you: that is completely wrong. OPS does not run on mainframes, it runs on anything from NT to large UNIX boxes. And there are no management overheads whatsoever with OPS. There are however overheads, but they got absolutely nothing to do with management. Unlike the mainframes.

>
>Also, part of the answer is switched gig ethernet & keeping the # of
>physical nodes, relative to data volume & traffic patterns, optimal.
>That is, that the overhead of internode work isn't disproportionate to
>the volume of data. Back to decent colocation, building a pmap if you
>have to.

And if you can get me a switched gig ethernet totally dedicated to OPS, I'll run rings around the shared-nothing approach! This to say: if you throw hardware at a particular solution to solve an inherent design problem, then to be fair you should do the same to the other solution, just to see what happens. Otherwise the TCO thing goes out the window!

>
>Yes a EEE cluster *is* more complex to manage from one standpoint, but
>OTOH, Oracle OPS is reknown as being more complex to tune otherwise.
>Pick your poisons.

That is quite true. We're told that 9i solves that problem. I dunno, I'd like to see it first.

>
>To be completely honest, the greatest problem I see is the increasing
>chance of logical inconsistency across nodes during recoveries as the
>size of the cluster increases in a busy OLTP environment. Since the risk
>is hard to assess, I can only imagine that the risk is higher than I
>wish it were otherwise.
>

Interesting point. Again, something that doesn't happen with OPS.

Cheers
Nuno Souto
nsouto_at_optushome.com.au.nospam Received on Fri Aug 17 2001 - 05:54:54 CDT