What actual hardware contains the shares? If by chance it is a real Netapp (or IBM N-series), you can use snapdiff, which is the best possible solution.
If it's some other brand of NAS, you are stuck. And yes, it's dirt slow. I've done this at many customers, and CIFS is just a very slow protocol, dunno why. One thing that will help, add to the dsm.opt: resourceutilization 10 That way the client will start 4 pairs of sessions concurrently (each pair is 1 producer and 1 consumer thread, 4 concurrent pairs is the most you can get). I usually find it works much better running parallel. (It appears not to be a problem of how much data is being sent across the network, but a problem of traversing the directory trees via CIFS being very, very slow.) If there are a zillion directory entries on these shares, and if the client you are using is 32 bit or only has 4-8G of RAM, try adding also MEMORYEFFICIENTBACKUP DISKCACHEMETHOD That lets the client write the huge active file lists to temp workspace instead of keeping it in memory (and you may have to do that anyway when you go to resourceutilization 10 or the scheduler is likely to crash on a 34bit machine). You would think it wouldn't help, but sometimes it does. This suggestion isn't about speed but manageability: Take a look at setting up these schedulers using a proxy name instead of their own hostnames, and do GRANT PROXYNODE, so that all the filespaces are backed up to a common nodename that is different from the hostname (make it "NASSHARES", for example). That way when you go to do a restore, you'll always go to one place to look for the filespaces. What is likely to happen is that over time some of the shares will grow, and you may need to redistribute them across 3 proxy machines instead of 2, then nobody can remember where to look when starting a restore, etc. I have a customer where we have 4 proxy machines to back up 15 EMC shares. (We don't dedicate machines as proxies - they have their own functions, but minimal backup needs - one is a backup domain controller, one is a Vcenter server - in other words, machines that don't have much else to do during backup hours.) Each proxy machine has a scheduler service installed to run its own backup, and each has a scheduler service that runs a proxied backup for the EMC shares under the name NASHOME. When the helpdesk needs to do a restore, they always go to the same NASHOME machine to do the restore, and all the backups are visible and restorable from there. The TSM Admin can move, shift, add, subtract shares to the proxy machine backups, without confusing everybody who might need to do a restore. Proxy function very useful for that. W -----Original Message----- From: ADSM: Dist Stor Manager [mailto:ADSM-L@VM.MARIST.EDU] On Behalf Of Leonard, Matthew Sent: Sunday, March 16, 2014 2:23 PM To: ADSM-L@VM.MARIST.EDU Subject: [ADSM-L] best/most effecient way to backup a CIFS Share I'm trying to backup multiple CIFS shares and am trying to figure out the best way. This is what we did and I would like to know if there is a better way. We are using TSM 6.2 with a VTL. We setup two servers SXDCTPM and SXDCTPM01 and both are registered with Tivoli as nodes and assigned to daily schedules. The options file is below for one of the servers. NODENAME sxdctpm01 TCPSERVERADDRESS sxdctsm01 schedlogretention 7 ERRORLOGRETENTION 7 compression no COMPRESSALWAYS no schedmode prompted PASSWORDACCESS GENERATE exclude c:\...\* include \\sxdcfs02\P2\...\* include \\sxdcfs02\PACFinance\...\* include \\sxdcfs02\PolarShare\...\* include \\sxdcfs02\Public3\...\* include \\sxdcfs02\PublicShare2\...\* include \\sxdcfs02\Shared\...\* include \\sxdcfs02\TRAX$\...\* include \\sxdcfs02\XenTSRedir$\...\* DOMAIN ALL-LOCAL DOMAIN "\\sxdcfs02\P2" DOMAIN "\\sxdcfs02\PACFinance" DOMAIN "\\sxdcfs02\PolarShare" DOMAIN "\\sxdcfs02\Public3" DOMAIN "\\sxdcfs02\PublicShare2" DOMAIN "\\sxdcfs02\Shared" DOMAIN "\\sxdcfs02\TRAX$" DOMAIN \\sxdcfs02\XenTSRedir$ Is this the best way to backup CIFS shares. We are getting horrible performance and the backups seem to run forever. Is there a better way? Thanks Regards, Matthew J. Leonard Network Infrastructure Administrator IT Network Operations AtlasAir, Inc. matthew.leon...@atlasair.com<mailto:matthew.leon...@atlasair.com> 914-701-8042
