[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: NFS recommendation



On Sun, 06 Jun 1999 01:52:42 +0300 (IDDT), Ariel Biener
<ariel@fireball.tau.ac.il> wrote:

>On Sat, 5 Jun 1999, Udi Finkelstein wrote:
>
>
>  Hi,
>
>> >  From my experience, it would be a "BAD" idea to tie up between the
>> >system partitions and this partition you want to make available to Solaris
>> >clients.
>> Obviously, the data to be shared will be on a different partition.
>
>Better yet to have the `/' and swap and `/usr' on a completely different
>disk, and not on any of the 3 22GB disks. 
Not im my case. The 3 disk limit comes from the 4 IDE device limit (minus 1
for the CDROM). I don't want to add more disks now, as I'm trying to keep the
system cheap. If I'll need more disks, I'll add a Promise card.

>> The question is, is RH5.2 that bad? Is RH6.0 stable enough?
>> If performance goes up from 6MB/s to 8MB/s,I'de stay with 6MB/s, if it's
>> more stable. If performance is 0.6MB/s, that's a different story...
>
>There is alot more than just raw performance to this claim. RH 6.0 is
>stable, IMHO more stable than 5.2. 
So the first release of a 2.2 kernel is more stable than the 3rd release of a
2.0 based kernel? Usually the version X.0 syndrome dictates otherwise ;-)

>I would hardly even attempt to compare the NetApp WAFL file system, and
>the NetApp RAID with what you are going to offer. The ATA/33 IDE
>controllers and the Linux ext2fs are hardly any match to what NetApp
>offers. The avg. seek you get on NetApp due to the nature of WAFL is ~2ms.
>The way it writes and reads gives you ALOT of performance improvement on
>read/write opperations. So, due to inferior hardware (IDE) and filesystem
>(ext2fs), at least try to add some speed (ATA/66).

I'm sure you are right, but the cost of a single 18GB disk for the NetApp is
on the order of $3000, which is about what I plan paying for the complete NFS
server.

Actually, price wasn't really the problem with the NetApp. The problem was
that disk space was filling at a faster rate than our disk-buying rate, and
managing backups has become a nightmare (even with a DLT drive). Now we'll
separate the non-important files from the important files, and backup only the
NetApp.

Besides, I believe that the combination of NetApp for important source code
with a cheap Linux server for compilation object files, on a separate 100BaseT
link to a switching hub will be faster than the NetApp alone.

>--Ariel

bye,
udi