Are you considering using the X4500 as a disk-cache? Or are you simply 
trying to back the X4500 up?

I've been doing some testing with the X4500 as a disk-cache storage node 
and have found the following

* configure 1 big zpool (RAIDZ/RAIDZ2/Mirror, depending on your 
space/perf/reliability business needs)
* use adv_file as your device type
* configure multiple ZFS filesystems, Legato spawns 1 nsrmmd writer 
process per "device" or adv_file type. If you only configure 1 adv_file 
device, you only get a single writer, no matter how much data your 
clients throw at it. Limits you to ~200-250MB/sec. Set reasonable limits 
on the number of concurrent streams each device will handle to load 
balance across all available devices.

ie. 8 clients sending 2 streams each. Configure 4 ZFS filesystems and 
let each filesystem handle 4 concurrent streams.

Also, configuring multiple adv_file devices gives you the ability to do 
multiple clone sessions if that is part of your process. So if you have 
4 tape drives waiting, 4 adv_file devices will be able to run 
simultaneous clones.

I haven't found any advantage in changing the recordsize. I have not 
tested compression yet. You can tune the NSR BLOCK SIZE, but I've found 
the default to be robust. Anything much over 256k has shown to only 
cause higher service times and wait queues on the disks, with no 
increase in streaming performance (this was tested more on NBU though).

Hope that gives you some starting points

-ryan




Itay Menahem wrote:

>Hi,
>Are there any Legato backup system considerations that I should take in 
>account while building zpools and file systems on a Sun X4500 machine?  I was 
>thinking of such consideration such as the zpool size, file system properties 
>such as compressions,  record size etc.
>As far as I understand EMC Legato NetWorker 7.3.2 supports Sun solaris ZFS.
>
>thanks,
>
>Itay.
> 
> 
>This message posted from opensolaris.org
>_______________________________________________
>zfs-discuss mailing list
>zfs-discuss@opensolaris.org
>http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
>  
>

_______________________________________________
zfs-discuss mailing list
zfs-discuss@opensolaris.org
http://mail.opensolaris.org/mailman/listinfo/zfs-discuss

Reply via email to