OS: Solaris 8 (Sparc)
NW: 7.1.3 (Sun EBS variant) /w AlphaStor (Latest 2.x I think?)
Server: E450 /w 4x480Mhz CPU and a bunch of internal disks (mostly /nsr).
S-Node: V240 /w 2x1.2Ghz CPU (Also our AlphaStor server)
Our DBO license is Tier 5 (No Disk Limits).
The 60 TB is presented from a FlexStore RAID unit as multiple LUNs less than
1 TB, and used by Veritas VxVM as part of normal Concat volumes (the
underlying RAID device is a series of 3+1 and 4+1 RAID-5 with several
hotspares). We use Veritas DMP to allow multiple paths instead of STK-RMAN,
and the array is direct-connected to our Server and Storage Node.
Largest adv_file is just under 16 TB (we hit some kind of limit in Veritas -
can't remember what right now). We have one each of Database, Full, and
Incr disk devices on our Server and Storage Node (each of approx equal
size), with space left over for a 'small,' special-purpose disk device on
each (splitting the load relatively well). About 30 TB each system.
There are staging policies in place for everything but the Full volumes,
since those so far have held our entire backup history with some room for
growth. Storage Node affinity per-client is used for (re)balancing as
needed, though I also have some general guidelines regarding what qualifies
for server or SN.
Since all the adv_file volumes are VxFS I'm not encountering any major
issues with the size that you would a UFS filesystem. ZFS was not available
at the time the disk originally came in (this is our second array - long
story, don't ask). So far performance has done well without needing to
resort to Veritas volume defrag, and even though the array takes a beating
the failures from that have been relatively few.
Right now our bottleneck is server I/O - mostly network, as we use gigabit -
but between the disk and LTO-2 drives, we're well within our windows again
to not worry as much about server hardware (other than the 450 being EOSL).
This current RAID unit is stable enough that I've considered placing /nsr on
it as a VxFS volume. Since the format is endian-neutral, upgrading from
SPARC to x64 (i.e. 4600) should be relatively simple then (but that's
another discussion).
--TSK
On Thu, 12 Apr 2007 07:52:05 -0400, Teresa Biehler <tpbsys AT RIT DOT EDU>
wrote:
>How are you slicing up the 60TB? What is your largest adv_file device?
>How do you balance your backups over the multiple adv_file devices
>(since NW will not span adv_file devices)?
>
>Also, you mention a 15TB adv_file device. Have you had any issues with
>file systems this large? Do you need to do any sort of file system
>maintenance (defrags) to maintain performance? What OS and version of
>NW are you using?
>
>Thanks.
>Teresa
To sign off this list, send email to listserv AT listserv.temple DOT edu and
type "signoff networker" in the body of the email. Please write to
networker-request AT listserv.temple DOT edu if you have any problems with this
list. You can access the archives at
http://listserv.temple.edu/archives/networker.html or
via RSS at http://listserv.temple.edu/cgi-bin/wa?RSS&L=NETWORKER
|
