Re: Oracle Performance on RAID 5

Sure, but the point I was trying to make is 'suffer' compared to what?

I'm saying that you can have excellent disk performance compared to a RAID5 array (if that's what you want) by considering alternative configurations (some of which I pointed out). Also, putting everything on RAID5 may *cost in read performance* - what happens if your queries sort or aggregate on disk? They *write*, that's what. Likewise redo and rollback. No database just *reads*, period.

What you have to decide is *what read performance* you want, not whether RAID5 is good enough (as it may or may not be, and partially or fully implemented across your db). If you are saying you *have* a RAID5 array that works well in it's current configuration, then fine, no issue. But what your vendor has said is *partly (potentially)* true, RAID5 may not give you the best read performance, but then it may be no worse than single spindles. Likewise it might be much, much slower than the same spindles configured as a stripe set (and the same is true for RAW v non-RAW).

Basically you can't ask questions like 'has anyone noticed a drop', because it's meaningless because it has no context.

Please ignore this mail if it wasn't what you wanted to hear, as the last one seemed to annoy you some what. What you seemed to want based on you last post was for someone to say, 'No, I haven't noticed a drop', and yet no-one has said that, because really they can't. However if you want it, here goes: 'No, I haven't noticed a drop'. (But compared to what?.. And we have 1TB of storage across a mass of Oracle db's, some RAID5, and some not.)

MotoX.

>Robert Hart wrote in message <3588bdca.0_at_champ>...
>Hi

[snip]

>What concerns me is Vendor claims that I will suffer
>a performance hit on reads. All the literature I've read
>indicates that this is not true. My experience on RAID5
>has been good. No read performance degradation in
>my experience. I'm just asking if anyone has had
>a drop on read performance for reads.
>
>l8r
>Robert Hart
Received on Thu Jun 18 1998 - 12:16:31 CDT