Motherboard query...
Many of your questions may have already been answered in earlier discussions or in the FAQ. The search results page will indicate current discussions as well as past list serves, articles, and papers.
Jakob Østergaard jakob at unthought.netFri Mar 1 00:34:59 PST 2002
- Next message: Motherboard query...
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
On Thu, Feb 28, 2002 at 06:52:42PM -0800, Trent Piepho wrote:
> On Thu, 28 Feb 2002, Mark Hahn wrote:
> > > Raid 5 is costly. As I understand it (not very well) there's 3 writes for
> > > RAID 5. Reading can be slow, but not nearly as bad. Obviously you dont want
> >
> > I've never quite understood why people think R5 is expensive.
> > yes, you have to do the parity calculation, and a 1 byte write
> > does require an update of the parity block for that stripe,
> > but it's not egregious.
>
> In order to update a single RAID5 block you need to do two reads and two
> writes.
Wrong.
If you're writing not just one-bit-here-and-one-bit-there, but streaming
megabytes to the disk - then the buffer cache will already hold the information
needed for parity calculation (because userspace just told the kernel to write
it to the disks anyway). Thus, to write a full stripe including parity on an N
disk RAID-5 - you need *zero* reads, and N writes (one per disk) and they
parallelize and "pipeline" perfectly.
Of course this is slower than RAID-0 (because it *is* not as simple - there are
more lines of code - and the XOR being fast, still isn't free) - but it does
not have to be a lot slower.
> The two reads are the block to be written and the parity block. You
> need to XOR the old data block ^ old parity block ^ new data block, to get the
> new parity block. Then write the new parity and the new data. Much slower!
Yep - for a naive implementation. There are many performance tricks that can
be pulled, and Linux currently pulls some of them. Being smart about reads on
RAID-5 is one of them.
> For instance, the 3ware 7410 card we get something like 75Mb/sec writes to 4
> drives in raid 0 and 5.6MB/sec in raid 5.
Well, luckily one *implementation* with poor performance doesn't prove that efficient
implementations are impossible :)
('implementation' here meaning both software and hardware)
Cheers,
--
................................................................
: jakob at unthought.net : And I see the elder races, :
:.........................: putrid forms of man :
: Jakob Østergaard : See him rise and claim the earth, :
: OZ9ABN : his downfall is at hand. :
:.........................:............{Konkhra}...............:
- Next message: Motherboard query...
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
More information about the Beowulf mailing list
