Re: Your experience with HP EVA 8100
Date: Mon, 02 Nov 2009 13:23:18 +0200
I'm having experience with EVA storages starting with 1st generation (EVA3000), 2nd and now 3rd generation. We are having both installations with UNIX (Linux or HP-UX) and Windows. In all these installation the bottleneck was never in the storage, especially with EVA8100 which is the business critical storage of the mid-range class and it has 8 host and device ports per controller pair. As you probably know the maximum performance with the EVA storages can be achieved with FC disks when you are using maximum drives in one disk group, because reads and writes are executed at once in parallel at all disks. Apart from architectural design these are few thinks to consider regarding (if any) bad performance: EVA firmware, SAN switches firmware, HBA driver version, multipath software, load balancing policy! and file system block size.
Martin Bach wrote:
> Hi listers,
> I was wondering what your experience with the EVA 8100 was.
> The background: I migrated a 2 node 10.2.0.3 32bit system to (a lot)
> more powerful hardware with plenty more of everything. Now the
> bottleneck shifted, from db server to storage which I anticipated.
> Example: the backup to SAN storage using 2 channels (1.5T database + a
> few hundred gb of archived logs) always took around 6 hours on the old
> hardware but it seems the same script completely saturates the
> writeback cache of the array. Apparently there have been 250 busy
> commands in a period of 10 minutes. I don't think that a 2 channel
> backup should do this.
> On top of that it seems that there isn't an awful lot of tuneables on
> the array but that could be my lack of understanding of the thing.
> So how do you like the array?
> Martin Bach
> OCM 10g