RP5470 PERFORMACE

[JOHN_595]

I am trying to do a comparison between the systems listed below;

1. two HP RP5470 servers with the following config;
4 X 750 MHz CPUs, 8GB RAM, 4 X 36GB 15K rpm LVD , connected to a Storage server 7110 with 15 X 73GB FC HDD.

2. two HP Proliant ML 570-G2 servers with the following config;
4 X Xeon 2GHz CPUs, 8GB RAM, 4 X 73 15K HDD, connected to MSA 1000 storage unit with 14 X 73GB HDD.

the two above solutions to be clustered servers ; However need to know which will give better performance for using HPUX 11i and Oracle 10G.
Are there any direct comparison, that I can check as a result, ie, a chart, table,...
Lastly, What Application server is recommended to be used in conjungtion of my above solution?

[Jeff Schussele]

Hi John,

Well it will simply be option #1 as the ProLiant can't run HP-UX - just Winblows & Linux. The rp5470 will run HP-UX 11i and Oracle 10G.
But I'm curious as to why you'd use 10G on a single system. It's designed for grid computing which implies multiple servers. If this is to be a single system solution I'd think 9i would be a better solution simply for no other reasons then cost.

My 2 cents,
Jeff

[Alzhy]

If you're thinking of 10G RAC implementation..

I would think the Proliant Implementation running Linux or Solaris 10 (you cannot run HP-UX, btw.) will blow the rp5470 away. The clock speed alone will mean Oracle background processes will execute faster. I know DB processing is not all about clock speed but certain X86 implementations (with the proper OS and tuning) most of the time just blow their RISC counterparts away.

Or benchmark yourself if you've the resources.

[JOHN_595]

OOPS?
I meant to just consider it as a performance comparison, and surely I would have run HPUX and oracle on HP RP5470 and Linux and Oracle on the Proliant platform. I have attached a link which I would appreciate your comments on them.

http://www.oracle.com/apps_benchmark/html/0319B_Report1.html

http://www.oracle.com/apps_benchmark/html/0147B_Report1.html

[Alzhy]

Well John The benchmarks speak for themselves.

[Fred Ruffet]

Be carefull with benchmarks. Test they perform often do not correspond to real things you'll do on your servers.

4 CPUs will probably be better than 2 when having lots of sessions. If you have lots of IOs, disks perfs are importants. Those things are not really revealed by benchs.

They also don't include into their benchmarks your skills on each OS and support you'll have.

Regards,

Fred