11-28-2012 04:07 AM
Hi everyone,
I have an environment with Netbackup 7.5.0.3 on Windows 2008 that currently makes backup of several NetApp NAS to VTL using the NDMP protocol.
We are looking to make improvements in the backup/restore performance of those. I made a test for a backup using the Netbackup Accelerator technology on a CIFS share without any improvement. Details in the jobs log showed that accelerator was used, on first and following backups, but everything was copied again and no time improvement was observed.
I have found on the NetApp Documentation the following:
Can anyone confirm that Accelerator can not be used on these ???
What would be the proper method to improve backup / restore performance over the NDMP protocol?
Thank you in advance,
Amer.
11-28-2012 06:03 AM
Support for NAS CIFS shares is documented in this blog:
https://www-secure.symantec.com/connect/blogs/plan-better-backup-break-backup-window-netbackup-accelerator
Extract:
There are a number of great use cases for this feature today such as large file systems, NAS, and remote offices. It can dramatically improve the backup of large network attached storage (NAS) filers, using NFS or CIFS, rather than remote NDMP, because only the changed files are backed up.
You will not have the option to enable 'Change Journal'. So, the CIFS share will backed up similar to a Unix/Linux file system backup where this feature is not available.
11-28-2012 06:17 AM
Accellerator should be OK to do this although as far as i am aware tests have been done using NFS rather than CIFS
The change journal just makes it even faster - accelerator keeps its own track log so is not dependant on the change journal
The first backup may take a lot longer than an NDMP backup but after that it should be better
If the second run (with accelerator enabled in the schedule and policy attributes) backed up the same amount of data then it does sound like something was not quite right - can you post the full details of the first and second jobs logs for us to see please
11-28-2012 07:09 AM
These are the logs from two test backup jobs backed to a MSDP
500K files 289GB size backup
FIRST BACKUP JOB
125 Backup Done 0 TST_CPR01NAS004 full v0023bkp002 v0023bkp002 29/05/2012 17:00:00 STU_MSDP_V0023BKP002_S 30/05/2012 1:16:51 1 289635287 551546 08:16:51 9729 29/05/2012 17:00:00 08:16:51 28%
29/05/2012 17:00:00 - Info nbjm(pid=5996) starting backup job (jobid=125) for client v0023bkp002, policy TST_CPR01NAS004, schedule full
29/05/2012 17:00:00 - Info nbjm(pid=5996) requesting STANDARD_RESOURCE resources from RB for backup job (jobid=125, request id:{6719FFA5-4D1C-418A-BFC8-FA65AACD4E74})
29/05/2012 17:00:00 - requesting resource STU_MSDP_V0023BKP002_S
29/05/2012 17:00:00 - requesting resource v0023bkp002.NBU_CLIENT.MAXJOBS.v0023bkp002
29/05/2012 17:00:00 - requesting resource v0023bkp002.NBU_POLICY.MAXJOBS.TST_CPR01NAS004
29/05/2012 17:00:00 - granted resource v0023bkp002.NBU_CLIENT.MAXJOBS.v0023bkp002
29/05/2012 17:00:00 - granted resource v0023bkp002.NBU_POLICY.MAXJOBS.TST_CPR01NAS004
29/05/2012 17:00:00 - granted resource MediaID=@aaaaf;DiskVolume=PureDiskVolume;DiskPool=MSDP_V0023BKP002_S;Path=PureDiskVolume;StorageServer=v0023bkp002;MediaServer=v0023bkp002
29/05/2012 17:00:00 - granted resource STU_MSDP_V0023BKP002_S
29/05/2012 17:00:00 - estimated 110778 Kbytes needed
29/05/2012 17:00:00 - Info nbjm(pid=5996) started backup (backupid=v0023bkp002_1338303600) job for client v0023bkp002, policy TST_CPR01NAS004, schedule full on storage unit STU_MSDP_V0023BKP002_S
29/05/2012 17:00:00 - started process bpbrm (3672)
29/05/2012 17:00:01 - Info bpbrm(pid=3672) v0023bkp002 is the host to backup data from
29/05/2012 17:00:01 - Info bpbrm(pid=3672) reading file list from client
29/05/2012 17:00:01 - Info bpbrm(pid=3672) accelerator enabled
29/05/2012 17:00:01 - Info bpbrm(pid=3672) There is no complete backup image match with track journal, a regular full backup will be performed.
29/05/2012 17:00:01 - Info bpbrm(pid=3672) starting bpbkar32 on client
29/05/2012 17:00:01 - connecting
29/05/2012 17:00:01 - connected; connect time: 00:00:00
29/05/2012 17:00:02 - Info bpbkar32(pid=7328) Backup started
29/05/2012 17:00:02 - Info bptm(pid=7364) start
29/05/2012 17:00:02 - Info bptm(pid=7364) using 262144 data buffer size
29/05/2012 17:00:02 - Info bptm(pid=7364) setting receive network buffer to 1049600 bytes
29/05/2012 17:00:02 - Info bptm(pid=7364) using 30 data buffers
29/05/2012 17:00:04 - Info bptm(pid=7364) start backup
29/05/2012 17:00:06 - begin writing
30/05/2012 1:16:31 - Info bpbkar32(pid=7328) accelerator sent 296586533888 bytes out of 296586533888 bytes to server, optimization 0.0%
30/05/2012 1:16:31 - Info bpbkar32(pid=7328) bpbkar waited 124 times for empty buffer, delayed 929 times.
30/05/2012 1:16:31 - Info bptm(pid=7364) waited for full buffer 569692 times, delayed 1812880 times
30/05/2012 1:16:37 - Info bptm(pid=7364) EXITING with status 0 <----------
30/05/2012 1:16:37 - Info v0023bkp002(pid=7364) StorageServer=PureDisk:v0023bkp002; Report=PDDO Stats for (v0023bkp002): scanned: 289651453 KB, CR sent: 207616905 KB, CR sent over FC: 0 KB, dedup: 28.3%
30/05/2012 1:16:38 - Info bpbrm(pid=3672) validating image for client v0023bkp002
30/05/2012 1:16:41 - Info bpbkar32(pid=7328) done. status: 0: the requested operation was successfully completed
30/05/2012 1:16:41 - end writing; write time: 08:16:35
the requested operation was successfully completed(0)
SECOND JOB, NEXT DAY.
132 Backup Done 0 TST_CPR01NAS004 full v0023bkp002 v0023bkp002 30/05/2012 17:00:01 STU_MSDP_V0023BKP002_S 31/05/2012 1:14:45 1 289975565 551964 08:14:44 9780 30/05/2012 17:00:01 08:14:44 99%
30/05/2012 17:00:01 - Info nbjm(pid=7264) starting backup job (jobid=132) for client v0023bkp002, policy TST_CPR01NAS004, schedule full
30/05/2012 17:00:01 - Info nbjm(pid=7264) requesting STANDARD_RESOURCE resources from RB for backup job (jobid=132, request id:{9BC79728-A332-4C8C-81AB-E88D413C64FA})
30/05/2012 17:00:01 - requesting resource STU_MSDP_V0023BKP002_S
30/05/2012 17:00:01 - requesting resource v0023bkp002.NBU_CLIENT.MAXJOBS.v0023bkp002
30/05/2012 17:00:01 - requesting resource v0023bkp002.NBU_POLICY.MAXJOBS.TST_CPR01NAS004
30/05/2012 17:00:01 - granted resource v0023bkp002.NBU_CLIENT.MAXJOBS.v0023bkp002
30/05/2012 17:00:01 - granted resource v0023bkp002.NBU_POLICY.MAXJOBS.TST_CPR01NAS004
30/05/2012 17:00:01 - granted resource MediaID=@aaaaf;DiskVolume=PureDiskVolume;DiskPool=MSDP_V0023BKP002_S;Path=PureDiskVolume;StorageServer=v0023bkp002;MediaServer=v0023bkp002
30/05/2012 17:00:01 - granted resource STU_MSDP_V0023BKP002_S
30/05/2012 17:00:01 - estimated 551354 Kbytes needed
30/05/2012 17:00:01 - Info nbjm(pid=7264) started backup (backupid=v0023bkp002_1338390001) job for client v0023bkp002, policy TST_CPR01NAS004, schedule full on storage unit STU_MSDP_V0023BKP002_S
30/05/2012 17:00:02 - Info bpbrm(pid=2628) v0023bkp002 is the host to backup data from
30/05/2012 17:00:02 - Info bpbrm(pid=2628) reading file list from client
30/05/2012 17:00:02 - Info bpbrm(pid=2628) accelerator enabled
30/05/2012 17:00:02 - Info bpbrm(pid=2628) There is no complete backup image match with track journal, a regular full backup will be performed.
30/05/2012 17:00:02 - Info bpbrm(pid=2628) starting bpbkar32 on client
30/05/2012 17:00:02 - started process bpbrm (2628)
30/05/2012 17:00:02 - connecting
30/05/2012 17:00:02 - connected; connect time: 00:00:00
30/05/2012 17:00:05 - Info bpbkar32(pid=2156) Backup started
30/05/2012 17:00:05 - Info bptm(pid=4736) start
30/05/2012 17:00:05 - Info bptm(pid=4736) using 262144 data buffer size
30/05/2012 17:00:05 - Info bptm(pid=4736) setting receive network buffer to 1049600 bytes
30/05/2012 17:00:05 - Info bptm(pid=4736) using 30 data buffers
30/05/2012 17:00:07 - Info bptm(pid=4736) start backup
30/05/2012 17:00:11 - begin writing
31/05/2012 1:14:35 - Info bpbkar32(pid=2156) accelerator sent 296934978560 bytes out of 296934978560 bytes to server, optimization 0.0%
31/05/2012 1:14:35 - Info bpbkar32(pid=2156) bpbkar waited 23 times for empty buffer, delayed 321 times.
31/05/2012 1:14:35 - Info bptm(pid=4736) waited for full buffer 603602 times, delayed 1874792 times
31/05/2012 1:14:39 - Info bptm(pid=4736) EXITING with status 0 <----------
31/05/2012 1:14:40 - Info v0023bkp002(pid=4736) StorageServer=PureDisk:v0023bkp002; Report=PDDO Stats for (v0023bkp002): scanned: 289991743 KB, CR sent: 2560280 KB, CR sent over FC: 0 KB, dedup: 99.1%
31/05/2012 1:14:40 - Info bpbrm(pid=2628) validating image for client v0023bkp002
31/05/2012 1:14:41 - Info bpbkar32(pid=2156) done. status: 0: the requested operation was successfully completed
31/05/2012 1:14:41 - end writing; write time: 08:14:30
the requested operation was successfully completed(0)
11-28-2012 07:10 AM
Thanks for your answer. I posted the job details.
11-28-2012 07:12 AM
Thank you Marianne for you quick response. I posted the job details in case you want to take a look.
11-28-2012 07:53 AM
Take a look on the client (media server?) under:
<installdir>\NetBackup\track\<masterserver>\<storageserver>\<client>\<policy_name>\
and see what folders there are in there - these should specif the backup specification
If they are different then it has not detected thetrack logs correctly
Or maybe your media server is short of disk space and it has not had the space to create the track log though it should only actually be about 150MB based on your data
The key should be in that track log area
Hope this helps
11-29-2012 12:15 AM
Hi Mark,
Thank you for your answer.
I've checked the track folder, and there is a folder structure under with the name of the policy. This folder is empty.
However there is another folder structure for another test policy I made, this time for a local backup, and I can see the track_journal.v1.dat files.
There is plenty of space in the disk.
Question would be how the NAS is going to communicate netbackup the track changes in its filesystem from the last backup if Netapp CIFS share does not support Logging of NT File System (NTFS) events in the change journal ?
Regards
11-29-2012 03:02 AM
The change journal is a Windows thing - Accelerator should not need that, only its own track log - but it does not look like the track files are being created for some reason.
It should work and is supported so it may be worth opening a case with Symantec of you are sure everything is setup correctly
02-27-2013 07:39 AM
Did you ever get Accelerator working with your NetApp CIFS share?
Is this actually supported and works? Is it necessary to map all your CIFS shares as drive letters to your media servers to accomplish this?
02-27-2013 08:59 AM
I have been trying this since the day Accelerator came out. NFS works fine, CIFS never worked, with Nbu Client service logon account or not.
It would seem that Accelerator does not track changes on CIFS UNC path locations. (Not talking about NTFS USN change journal, of course.)
I could only guess that this has something to do with CIFS being user connection session centric, and NFS being more computer centric (it "becomes" part of the host's "local" file system, to a degree).