Simpler Navigation for Servers and Operating Systems
Completed: a much simpler Servers and Operating Systems section of the Community. We combined many of the older boards, so you won't have to click through so many levels to get at the information you need. Check the consolidated boards here as many sub-forums are now single boards.
Showing results for 
Search instead for 
Did you mean: 

Oracle8i performance on SuperDome and XP512

Go to solution
Honored Contributor

Oracle8i performance on SuperDome and XP512

Hello Database Admins,

if you want to play a little with a SuperDome server and a XP512 DiskArray, you just have to come to hp russia, it seems ;-)

Right now, I do have a chance to do some high-end benchmarking on that combination, so I would like to get your experience on that!

What do think (or even know) about to tune this beast? 16 CPUs, 16 GB of RAM, 2 FC-links to XP512, 5 ParityGroups of the fast 18 GB drives (but in RAID5), 64bit Oracle 8.1.6, whole nights to toy around with :-)

I am thinking about trying
- about 10GB of SGA in 1GB chunks to "stripe" RAM over CPU-blocks
- POPS with 256MB pages for data (10 GB of SGA)
- timeslice=11 or 12
- async I/O (yes, this is dangerous)
- Veritas VM to do LoadBalancing on the FC links
- Raw-Devices for Oracle, of course (data, and redo-logs)
- No-ArchiveLog-Mode (dangerous, too)
- High Performance Mode for the CHIP used
- Using a Brocade FC-Switch to actually use 4 CHIPs over the switch to connect to the 2 FC-links from the SuperDome

Sadly, *my* XP512 is a little bit *narrow* on the CacheRAM side, so using CacheLUN for the redo-logs is not an option :-(

So, Ladies and Gentlemen, what is missing?
Just tell me, please, as I will summarize (as we used to say in the *good ol' days* of HPADM).
HP-UX? Disk-I/O? Oracle8i? XP512? SAN? else?

hey, I am quite curious, as I did not expect this to be possible here (yes, I have seen the Kremlin and the Red Square, already :-) and spending evenings/nights that way could be at least, say, entertaining? (the hotel is just 5m away).

Quite curiuos,
harry d brown jr
Honored Contributor

Re: Oracle8i performance on SuperDome and XP512


After not being able to answer your question on the high-performance CHIPS mode on the XP512, what makes you think we can answer anything "else"?

Sounds like you are really enjoying yourself over there!


live free or die

Live Free or Die
Honored Contributor

Re: Oracle8i performance on SuperDome and XP512

Hello Harry,

what are you doing there - an answer within a minute, or so?
Working hard on your "hat" it seems :-)
Understanding this as your "moral support" this is worth at least a single point, is it not?

And for the "rest" of you: one can always hope!

Waiting for more answers,
harry d brown jr
Honored Contributor

Re: Oracle8i performance on SuperDome and XP512

Did you ever find an answer to your High-Performance CHIPS question? I didn't find much on it, but I sure did learn a lot about the XP512 - probably more than I needed too :-))

I usually have a lot of time in between building virtual partitions on an N-class (6 partitions). Good healthy fun!


live free or die
Live Free or Die

Re: Oracle8i performance on SuperDome and XP512

A couple of additional questions;
How much cache on the XP? Over how many boards?
How much memory for the XP?
How many ACP's do you have in the frame?
How many drives do you have? 20? How have the drives being split up?
Do you have both left and right cabinets? How many on both side? I am guessing Right 1 only?
On the host, do you have AutoPath XP?
You have 4 CHiPs running in High Performance mode. Are all LUN's visable through all CHiPs?
How has the SAN being configured? Any Fiber Zone's? Any Volume Logix?

What sort of expansion are they planning on this box? A SuperDome with 16CPU's, 16Gb memory and only 270Gb of storage!

Honored Contributor

Re: Oracle8i performance on SuperDome and XP512

Hello Andrew,

well, actually I do not know why the SuperDome is here :-) the XP is just needed for a training...
So we kind of took the chance of "using" it...
And for your other questions:
- 1GB Cache RAM
- 2 ACP pairs
- 10 drives, 18GB in RAID5
- DKC and DKU R1 only
- no AutoPath, no AutoLUN, no CacheLUN
- 2 CHIP pairs
- SAN: XP - Brocade SW - SD
still working to set the CHIPs to HP-mode - we had duplicate IP-addresses with the XP/SVP, seems we are on a not-so-private-LAN here...
Hopefully that accounts for all those problems we have experienced.
Honored Contributor

Re: Oracle8i performance on SuperDome and XP512

Why only two fc links?
You've got 12 per/corecell?
How much memory in the dome?
Is it the same per cell?

How are you configuring your partition/partitions? Whats your cell assignment.. same XBC?

How are you planning your io to the xp in terms of your partitition.

Did you have a look at the full results for the hp test where this was done already? (except with fc60's)
You'll get kernel params there too.

It works for me (tm)
Thierry Poels_1
Honored Contributor

Re: Oracle8i performance on SuperDome and XP512


to me it seems you're only missing Oracle 9i to be on top of the world ;-)

All unix flavours are exactly the same . . . . . . . . . . for end users anyway.
Volker Borowski
Honored Contributor

Re: Oracle8i performance on SuperDome and XP512


an Oracle aspect:

Having 10GB of SGA with assumed big part of it for the databasebuffer and big update activity,
take care how you configure your online redo logs.
If they are too small, and switching around all the time, the database might have difficulties to manage the checkpoints.

Since you consider no_archive_mode, it might be well to enlage these as well, to have at least some stuff to recover from.
If they are too large, an instance recovery after a shutdown abort may take some time. So usally I go for several midsize online logs (consider even 10 to 20 groups of 50MB-80MB files) this balances the checkpoint stuff, because a log switch enables Oracle to process the first list of updated blocks, but does not block the logwriter as long as there are enough log groups to continue writing.

Another point of management: With the database in no_archive_mode, you will not be able to do online backups....(if application permits, or snapshotmounts (of the rawdevices) for offline backups possible, OK).

The thing you should watch out for is an alert-log-message stating "Checkpoint incomplete". Each time you have this, it means, the entire database waits for the dbwr to complete a checkpoint, before it permits the logwr to overwrite an online redo logfile.

But to be honest, I never had to deal with a machine like this. Do not know if this is relevant in this case.

Have fun with this toy.
Honored Contributor

Re: Oracle8i performance on SuperDome and XP512

just to add the doc link.

Oracle8 Enterprise Edition v8.1.7.1
TPC-C Throughput 197,024
a snap at 8,522,104 US $
256GB memory: 2.4M$!
The SD32000 was only 280k!
(only PA8600 and 48way SD)

I like the use of banks of the AutoRaids for swap space... wow!!!

It works for me (tm)
Sridhar Bhaskarla
Honored Contributor

Re: Oracle8i performance on SuperDome and XP512


Disk I/O is of the bigger concern in oracled systems.

Did you consider doing a striping at the system level?.

I have seen a huge difference between default setup (using the LUNs sequentially) and the stripe setup. I have created a graph here that shows different throughputs for different stripe sizes.

1. extent level striping with Extent size=4MB
2. LV stripe of 64MB
3. LV stripe of 128MB
4. LV strip of 256MB
5. LV stripe of 512MB
6. LV stripe of 1024 MB

You will definitely notice the difference.

We tried cache LUN. It didn't buy us much. Instead we can try Solidstate for redo logs.


You may be disappointed if you fail, but you are doomed if you don't try

Re: Oracle8i performance on SuperDome and XP512

Hi Wodisch,

2 CHiP pairs running High Performance mode will use up the ports on the XP, so the only suggestion is to make sure you stripe over all the pairs, and the 2 FC's on the HP.

On the XP the default stripe size for the Raid-5 is 48K, and with 5 Raid groups, you will probably want to use a stripe size of around 256K. We are still doing our testing, but the belief is that less than 256K is not advised, and once you go over 256K, there is not much difference. Our testing though is on a more populated XP.

I am guessing that you will have 4 LUN's per Raid-5 group, of about 12Gb's each. You will not really have enough spindles to put logs out on there own disks, so I would suggest the Oracle WP on SAME, for all LV's. Do a search for SAME (Stripe And Mirror Everything) on the website.

We have not used Veritas on HP yet as we are still at 11.0 as opposed to 11i. No comment available on Vertias therefore.

Re Oracle level, and Volker's comments, you can improve instance recovery speeds via the log_checkpoint_interval & timeout parameters, even if you have larger log files. BTW - there is no indication of transactional throughput.

Also, you should be using multiple members on the redo logs. This will provide a small degree of comfort, but I would not run a production database without archivelog.

You say you are using 8i, is this exclusive or OPS mode? If OPS, and you can move to 8.1.7 then consider OPFS, available from the OTN download links. Provides failover capabilities for the database and application connections.
Honored Contributor

Re: Oracle8i performance on SuperDome and XP512


finally I have to admit that I have not got the results, Andrew was changing faster than I was able to get copies of his results, so I actually do not know what was the "winning" setup. I still suspect the VxVM to have spoiled some of the tips...
Anyway: Thank you all for helping!!!
Kind Regards,