On Thursday December 21, [EMAIL PROTECTED] wrote:
> Neil Brown ([EMAIL PROTECTED]) wrote on 21 December 2000 15:35:
> > There is more work to be done to bring raid5 performance upto the
> > level of 2.2+mingos-patches, but this is a first, large, step on the
> > way.
>
> Interesting. What's the performance increase? The previous graphs in
> you site show that 2.4 was already better than 2.2, and the only point
> to be improved was reducing the difference in reading between raid 0
> and raid 5. Are there any new graphs?
Maybe you miss-interpreted the graphs??
The graphs on my site- e.g.
http://www.cse.unsw.edu.au/~neilb/wiki/index.php?LinuxRaidTest
show performance in 2.4.0 to be very poor, (e.g. dbench througput of
around 24Meg/sec) and that 2.4 with this patch (called "nb2.4" in the
graphs) is substantially better (e.g. dbench throughput of 35 to 45
Meg/sec).
There are new graphs but you'll have to come to my WiP at
linux.conf.au to see then .... (well, maybe I'll put them on the web
too).
I've got sequential read performance up above 2.2 levels and
appropriately close to raid0 level (you cannot get the same throughput
as raid0 for the same number of drives because of the parity block
that the drive needs to skip).
I've also nearly figured out how to make sequential write performance
match or exceed 2.2 levels.
NeilBrown
-
To unsubscribe from this list: send the line "unsubscribe linux-raid" in
the body of a message to [EMAIL PROTECTED]
