|
Hi
I
*think* but I may have entirely misread the comment that point 4 is just wrong.
It sounds like you only have one raid 5 set (which is shared between 2
machines). If this is the case then the clustering has no effect on the fault
tolerance/performance of the raid 5 set. Or am i just on crack here?
Point
5. writing to the datafiles *may* be acceptable on a raid5 set (it just depends
what the volume of write activity is - does anyone have any idea what the IO
requirements will be? if so do the calculations), writing to redo/archive almost
certainly won't be. I'd hold out for Raid 1 for the logs and maybe compromise on
Raid 5 for the data.
as for
the last point, I'd also like to see the justification for raid5 is 3 times more
likely to suffer dataloss. I'm afraid that a) I don't think I believe it and b)
you've got the logs anyway :( OK just a) really.
corrections and clarifications welcome.
You
may also wish to look at James Morle's SANE SAN paper at www.oaktable.net it would appear to be
pertinent (and he does know whereof which he speaks)
Niall
|
Title: Message
- FW: SAN configuration for Banner Sam Bootsma
- Re: FW: SAN configuration for Banner Paul Baumgartel
- Re: FW: SAN configuration for Banner Paul Drake
- Re: FW: SAN configuration for Banner Jared . Still
- Re: FW: SAN configuration for Banner Mogens N�rgaard
- Re: FW: SAN configuration for Banner Paul Drake
- RE: FW: SAN configuration for Banner Niall Litchfield
- RE: SAN configuration for Banner Niall Litchfield
- RE: SAN configuration for Banner Sam Bootsma
- RE: FW: SAN configuration for Banner Jared . Still
- RE: FW: SAN configuration for Banner Matthew Zito
- RE: FW: SAN configuration for Banner Matthew Zito
- RE: FW: SAN configuration for Banner Paul Drake
