Re: Hitachi StorEdge 9900 Series with Oracle?

> Is anyone out there using a Hitachi StorEdge 9900 series storage system
with their Oracle
> database(s)?
>
> We are in the beginning stages of configuring our brand new Hitachi 9970
for use with
> our Oracle database environment and we would very much like to have the
benefit of
> other Oracle shops' experience with the 9900 series. If you're using that
type of hardware
> and would be willing to entertain some questions about your configuration,
please respond.
> We'd be very grateful!
>
> --Jon
>

We are using some Hitachi 9900 Boxes for a half year. I can very very recommend these storage system. Forget nearly all you have read about the disadvantage of RAID-5. Try only people, which know the architecture of HDS. Inside HDS 4 physical disks build a RAID-5 array. There are special hardware for the calculation of the parity... There are spare disks which automaticly replaces failed disks (logicaly). It is not true, that You have a performance degration for a lot of hours, if a disks fails, but this don't occurs in our boxes. I believe, in the most cases You don't notice a failed disks in the application. We have 16 GB cache inside the array of 10TByte Boxes. The internal copy of 600GB inside the HDS for an internal snapshot (Shadow Image) was ready in nearly ONE hour. The disks are special disks with FC-AL on every disks with a higher MTBF as you normaly have. The storage system is optimized for RAID-5. So I can strongly recommend RAID-5.
For the creation of a database of 600 GB with only ONE FC-AL to HDS the traffic was 70 MB/second (stripe over the virtual disks from HDS and RAW devices) --> sequential write.
It is NOT TRUE that RAID1+0 is better for redo logs, because redo logs are written sequentially and a sequential I/O is very fine for RAID-5 on HDS and the cache.
Disadvantage of HDS:? The administration-interface, but if You have created Your LUNs You don't need it for the normal operation.

Think about some architecture features of HDS for performance: - Use 16 KByte Block-size (cache-unit 3 times 16 KB = 48 KByte   track size)
- Beacuse of 48 KByte track size try always a mulitple of 48KByte - Try a stripe size of 384 KByte (8 x track size, from internal   algorithms)

But You notice no performance degradation if not using above recomendations.

Put all files or RAW-devices on striped volumes (veritas volume manager ?) of the LUNs (every logical disk 14 GB). All on the same striped disks!! very easy administration whith no performance degradation.

I noticed, that the database creation on RAW devices was 100% faster, but other influences could be possible.

The HDS is the fasted storage subsystem, I ever seen.

Regards,
Eberhard Received on Wed Aug 21 2002 - 20:46:22 CDT