This is but a single data point, but if others can provide some 
additional data, it might help you triangulate what you want.

I'd also recommend 3ware over Areca - I've had 2 bad experiences with 
Areca involving data loss - only one of which I could really blame on 
the controller, but 3ware service is very good (as such things go).

Here's the bonnie++ output on a backup server I built for a lab:
This was on a very baseline box (single athlon64/.5GB RAM) outfitted 
with a single PCI-X 3ware 9550SX-12 controller driving 12 SATA-2, 
NQS-enabled 500GB WD disks in RAID5 with an XFS file system, 64K 
stripe, with readahead set to 16K (blockdev --setra 16385 /dev/sdb)

Email me if you want the exact spec.

Here's the bonnie++ output (it looks like it's going to be fragged by 
the line wrap tho - I can send you a file that will protect it and 
provide more details in  it as well).  It'll be cheaper of course now.  
The disks were on the 3ware approved list and were also among the 
cheapest available.  The SATA2 spec and the NQS makes a huge 
difference.  If you have to down-grade the disks to SATA1 (via 
jumper, because of controller incompatibility), performance drops to 
~1/3.

The salient number is 206298 KB/s for block writes and 543326 KB/s for 
block reads.  When I was doing some benchmarking on climate data 
reduction (netCDF files), I got real-world numbers close to what 
bonnie++ indicated.  Of course character performance is relatively 
awful.  And this is an XFS filesystem - as others noted, you'd 
probably want reiserfs (or maybe jfs?).  If you keep the journal on 
another controller, you should also get better performance.

Also note, this is with RAID5.  If I had really tried for maximum I/O 
with RAID 0 or 10, it should have been even higher.

Version  1.03       ------Sequential Output------ --Sequential 
Input- --Random-
                    -Per Chr- --Block-- -Rewrite- -Per 
Chr- --Block-- --Seeks--
Machine        Size K/sec %CP K/sec %CP K/sec %CP K/sec %CP 
K/sec %CP  /sec %CP
kstore       15000M 57150  96 206298  42 113941  29 63148  99 543326  
55 231.6   0
                    ------Sequential Create------ --------Random 
Create--------
                    -Create-- --Read--- -Delete-- -Create-- --Read--- -Delete--
              
files  /sec %CP  /sec %CP  /sec %CP  /sec %CP  /sec %CP  /sec %CP
                 16  6081  31 +++++ +++  5702  24  7243  35 +++++ +++  
1448   7
kstore,15000M,57150,96,206298,42,113941,29,63148,99,543326,55,231.6,0,16,6081,31,
+++++,+++,5702,24,7243,35,+++++,+++,1448,7

-- 
Harry Mangalam - Research Computing, NACS, E2148, Engineering Gateway, 
UC Irvine 92697  949 824 0084(o), 949 285 4487(c) 
[EMAIL PROTECTED]

-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys-and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
_______________________________________________
BackupPC-users mailing list
BackupPC-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/backuppc-users
http://backuppc.sourceforge.net/

Reply via email to