Re: Imported DB runs SLOWER on faster machine

Daniel Morgan <dmorgan_at_exesolutions.com> wrote:

>Either way ... it would take one heck of a horribly configured and tuned 4CPU
>Ultrasparc to be slower than a single CPU PC.
>
>Daniel Morgan

My moral condolence.

This machine was setup and configured by Sun people. I know them, I don't admire their capabilities, but they are not idiots. I myself had the suspicion that some trivial kernel setting or whatever would be causing the problem. I complained to Sun, but nothing.

You won't perhaps believe that right now I opened a case at Sun to investigate the miserable performance of the StorEdge A1000. They refused to engage with details and told me, if nothing is broken, they won't fix it. They won't recognize any performance test that I did and they won't do any test themselves until we prove what we say, (..which they won't recognize...).

In short: We are changing our Service Contract from Gold to Silber, stopped purchasing things from Sun and will switch to Linux on Intel based Servers very soon. I can't wait to get a Dell Rack Server with two Xeons (2.4 GHz) and really fast harddisks and lot of features for only about 10000$.

See my answer to the article from Sven Jensen in this same thread.

I don't know what could be so terribly wrong to cause such a slow speed. Well... wait... I DO know: At times, mostly during weekends, the server used to crash. After some investigation I found out, that the CPUs have a fabrication error that causes the built-in cache to corrupt (admited by Sun) when the CPUs are idle for a long time and data remain in the cache. Sun's solution: A patch that would regularly flush the cache every 10 sec. Sun told me, there is no performance impact from this solution... Do you believe them?

Go to www.spec.org and compare yourself.

Bye
Rick Received on Tue Jul 23 2002 - 16:47:12 CDT