> 3.8TB in 30hrs is the equivalent of 17MB/sec ?? Can someone verify this 
> number? If my math is correct, this is very slow for an NDMP drive, and I am 
> assuming that the drive is connected to the Celerra. i.e, the data is not 
> going through the LAN before being written to the NDMP drives. 
> 
> Assume that we are using NDMP configured drives that are connected directly 
> to the filer, I have seen LTO-3 drives can write (conservatively) 120MB/sec. 
> Again if my math is correct, then backing up 3.8TB should take approx 8.8 
> hours. 
> 
> If you can also backup 4 streams in parallel (Celerra allows for max of 4 pax 
> threads that allows for 4 backup streams), and assuming that each stream 
> backs up equal amount of data, then this should reduce the backup time to 
> approx 2.2 hours. In theory. 
> 
> So... is there 4 drives connected directly to your NS40, (scsi or fiber)? 
> 
> You mention that you do daily INC backups. Incremental backups are not 
> allowed with NDMP backups. You are only allowed to specify full or 
> differential type. If you specify incremental, then the filer will use 
> level=full, because the filer does not know what is incremental. 
> 
> Make sure you have Snapsure enabled on your Celerra, otherwise this can cause 
> your backups to fail. 
> 
> You generate multiple streams by spreading the data across multiple 
> partition. Not sure why you would not want to take advantage of this, 
> especially since you are trying to reduce the backup time. 
> 
> 
> Well, if you are limited to just one stream, then make sure that The NS40 is 
> writing directly to a locally connected drive to the Celerra. If possible, 
> replace the lto-3 with a faster drive lto-4, 5, 6... I know they are more 
> expensive, but so is the time needed to backup. 
> 
> Finally, see if EMC can do a health check on the NS40 to make sure it is 
> performing optimally. 


In my last meeting with our EMC crew we was lead to believe that given the size 
and number of files that the backups would be this slow normally, doesn't seem 
right to me, but they are the experts, right. EMC did run some health checks on 
the backup end but I will ask about the 480.
As far as the crashing part that is a comment from my team who has been here 
for years(I'm the new guy) I havent verified that. 

Everything is going Fiber from NS480 to tape drives. I am in the process of re 
organizing the groups so that the home folders get 3 hours of pretty much 
deticated use of the library, have a couple of small jobs run concurrently but 
they normally take about 15mins. I have 7 drives all fiber attached LTO3.
I am thinking about getting away from the NDMP for this folder so that I can 
use wildcards and use 26 saveset, that is what I did my other networker domain. 
 \home\a* ....
I would love to split the folders up but we now have 3 silos of Admins - 
design, operations and backup. Me being in the backup group fall on deaf ears 
when trying to get the other two to understand that even though it makes the 
helpdesk life easier to use 1 folder, it makes backing it up a real nightmare. 
Thanks for all the input so far, 
Ed Johnson

+----------------------------------------------------------------------
|This was sent by ed.johnson AT denvergov DOT org via Backup Central.
|Forward SPAM to abuse AT backupcentral DOT com.
+----------------------------------------------------------------------

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