On Thu, 30 Dec 1999, E Hunter wrote:
>
> I'm sure you'll let me know if these are better left to the sales
> people, but they will help me a great deal in completing the
> evaluation my company wants me to write.
Sales people will tend to paint a rosier picture than reality, but
the folks in here (generally being techies or engineering types) will
give you the straight goods. ;-) OTOH, sales reps are good for swag. :)
[Hint: to anyone reading from netapp... need more swag! Jackets,
mugs, blankets, shoulder bags, multitools, sunglasses, etc.? I've got
the shirts, the swiss army knife, the baseball cap and a few other
trinkets already. ;-) ]
> Basically, I understand from the literature I've read that the 760
> runs a sort of modified RAID level 4. In our particular environment,
> recoverability is not as important as speed,
The WAFL filesystem is extremely efficient when it comes to
writing out entire stripes (parity and all) to disk. I doubt there
would be any real performance gain assuming you switch the filer to
RAID 0 (which a Netapp can't do anyway). Filers win on access times,
not so much raw throughput. If you're looking for 100 MB/sec
sustained writes to disk, a Netapp isn't going to cut it (yet).
What sort of application will this be for? The other thing a
filer isn't fast at (and this is really the fault of NFS) is locking.
We have one particular financial application that locks/unlocks rows
in a database file like a sonofabitch, and it completely falls over on
an NFS filesystem, so we had to go back to local storage for it. Poor
software design if you ask me, but that's another story.
-- Brian Tao (BT300, taob@risc.org) "Though this be madness, yet there is method in't"
This archive was generated by hypermail 2b29 : Tue Jan 04 2000 - 10:20:56 EST