Backup Central

Has anyone tried restored an entire NetApp Volume with lots of files?
I'm currently attempting a test restore of our main groups data volume.
This volume is ~1.87TB in Size with ~3.9 million files.

We backup our NDMP volumes over the LAN using NDMP Remote Server (NRS) on our single MediaAgent to local disk attached to the mediaagent, then aux copy to tape.
Total time for the full backup job is ~30.5 Hours.

I'm attempting the restore from the second copy ( Aux Copy) on LTO-3 tapes directly to a new test volume on the netapp.
I did not select 'Use Direct Access Restore' as the help files says it will take longer for many files.

The restore job has been running for over 2 hours and says only ~7Gb of data has been transferred, which appears to be just the file and folder structure. The NetApp says only ~1.05 million files exist in this new volume, but the sizes of all the files are zero.

Should I expect that this full restore will take double the time of the backup phase? ~60 Hours?
Will the speed of the restore increase after the job builds the file/folder structure?

I've successfully restored individual files from an NDMP backup, but this is the first time that I'm attempting an entire volume restore.


Thanks in advance for any insight you can provide.

-Jon
__._,_.___
Reply to sender ([email]studer_je < at > hotmail.com?subject=Re%3A%20NDMP%20Restore%20progressing%20very%20slowly[/email]) | Reply to group ([email]commvault < at > yahoogroups.com?subject=Re%3A%20NDMP%20Restore%20progressing%20very%20slowly[/email]) | Reply via web post | Start a New Topic
Messages in this topic (1)
Recent Activity:

• New Members 6

Visit Your Group

Commvault Documentation here:
http://documentation.commvault.com/
Text-Only ([email]commvault-traditional < at > yahoogroups.com?subject=Change Delivery Format: Traditional[/email]), Daily Digest ([email]commvault-digest < at > yahoogroups.com?subject=Email Delivery: Digest[/email]) &bull; Unsubscribe ([email]commvault-unsubscribe < at > yahoogroups.com?subject=Unsubscribe[/email]) &bull; Terms of Use




.

[img]http://geo.yahoo.com/serv?s=97359714/grpId=7450851/grpspId=1707277520/msgId=4443/stime=1290454627/nc1=1/nc2=2/nc3=3[/img]
__,_._,___

If you are restoring the entire NetApp volume then the DAR option would have been automatically de-selected by the restore process. So not using the DAR option in this case is correct.

Also note you should not do anything with any of the 0 size files. The documentation may suggest un-sharing/un-exporting the volume when doing a restore to make sure nobody accesses any of the 0 size files.

The NetApp restore will first read and process a header at the start of the backup and layout the file system with all 0 size files. If your volume at that point in time has 1.05 million files and the backup of this volume has 3.9 million files then the file server likely is still laying out the file system. During this time there will not be a lot of data read from the backup (at least not a lot of bytes per hour) since the file server is spending it’s time preparing for the restore rather than actually restoring the file data. Once the file system has been complete created with 0 size files, the file server will begin to fill in the files – at that time the restore speed will increase. It is difficult to predict how long the restore will take since the restore time is highly data and file server dependent. You average file size is about 515k – which is not too small. Your restore will likely take more than 30 hours (the backup time). My guess is it will take less than 60, but be closer to 60 than to 30.

In Simpana 9.0 there is a backup set option for an image backupset – this option is available when using ONTAP 8.0 and later. With this option, block based full volume backups are used. Currently there is no granularity on restore – just full backups and full restores – no files are listed in the browse. The full volume restores are much faster – similar to the backup speed. The backups can be faster also. This manner of backup works well if full volume restores will be used.

Duncan












******************Legal Disclaimer***************************
"This communication may contain confidential and privileged material
for the sole use of the intended recipient. Any unauthorized review,
use or distribution by others is strictly prohibited. If you have
received the message in error, please advise the sender by reply
email and delete the message. Thank you."
****************************************************************

__._,_.___
Reply to sender ([email]prvs=094392d6ae=duncan < at > commvault.com?subject=Re%3A%20NDMP%20Restore%20progressing%20very%20slowly[/email]) | Reply to group ([email]commvault < at > yahoogroups.com?subject=Re%3A%20NDMP%20Restore%20progressing%20very%20slowly[/email]) | Reply via web post | Start a New Topic
Messages in this topic (2)
Recent Activity:

• New Members 4

Visit Your Group

Commvault Documentation here:
http://documentation.commvault.com/
Text-Only ([email]commvault-traditional < at > yahoogroups.com?subject=Change Delivery Format: Traditional[/email]), Daily Digest ([email]commvault-digest < at > yahoogroups.com?subject=Email Delivery: Digest[/email]) &bull; Unsubscribe ([email]commvault-unsubscribe < at > yahoogroups.com?subject=Unsubscribe[/email]) &bull; Terms of Use




.

[img]http://geo.yahoo.com/serv?s=97359714/grpId=7450851/grpspId=1707277520/msgId=4445/stime=1290526349/nc1=1/nc2=2/nc3=3[/img]
__,_._,___

Duncan,
Thanks for all the good info.

FYI, We are currently on Simpana 8 with all the latest patches/SP's installed. I'm looking forward to version 9.x when it becomes available to us.

It looks like the restore is progressing. ~250GB has been restored. It appears to be past the point where it lays out the file system and is actually restoring data.
Looking at the throughput of the tape drive transfer rate over the last 3 hours the data is being restored at ~14.5GB/hr. Based on that and the size of the backup being 1.87Tb, the estimated restore time is ~4.5 days.
This seems like a really long time for a restore.

We are working to break this volume out into smaller volumes, this may give us more motivation.

What's weird is that the backup job reports that the backup contained 3,936,197 objects. However, when I look at the source volume from NetApp's perspective, it shows 6.6 million files. The Restore testing volume also now shows 6.6 million files.

I'm wondering why the descrepancy in number of objects that CommVault displays, and number of current files that NetApp displays.

-Jon



________________________________
To: commvault < at > yahoogroups.com
From: duncan < at > commvault.com
Date: Tue, 23 Nov 2010 10:31:23 -0500
Subject: [commvault] Re: NDMP Restore progressing very slowly



If you are restoring the entire NetApp volume then the DAR option would
have been automatically de-selected by the restore process. So not
using the DAR option in this case is correct.

Also note you should not do anything with any of the 0 size files. The
documentation may suggest un-sharing/un-exporting the volume when doing
a restore to make sure nobody accesses any of the 0 size files.

The NetApp restore will first read and process a header at the start of
the backup and layout the file system with all 0 size files. If your
volume at that point in time has 1.05 million files and the backup of
this volume has 3.9 million files then the file server likely is still
laying out the file system. During this time there will not be a lot of
data read from the backup (at least not a lot of bytes per hour) since
the file server is spending it’s time preparing for the restore rather
than actually restoring the file data. Once the file system has been
complete created with 0 size files, the file server will begin to fill
in the files – at that time the restore speed will increase. It is
difficult to predict how long the restore will take since the restore
time is highly data and file server dependent. You average file size is
about 515k – which is not too small. Your restore will likely take more
than 30 hours (the backup time). My guess is it will take less than 60,
but be closer to 60 than to 30.

In Simpana 9.0 there is a backup set option for an image backupset –
this option is available when using ONTAP 8.0 and later. With this
option, block based full volume backups are used. Currently there is no
granularity on restore – just full backups and full restores – no files
are listed in the browse. The full volume restores are much faster –
similar to the backup speed. The backups can be faster also. This
manner of backup works well if full volume restores will be used.

Duncan











******************Legal Disclaimer***************************
"This communication may contain confidential and privileged material
for the sole use of the intended recipient. Any unauthorized review,
use or distribution by others is strictly prohibited. If you have
received the message in error, please advise the sender by reply
email and delete the message. Thank you."
****************************************************************



------------------------------------

Commvault Documentation here:
http://documentation.commvault.com/Yahoo! Groups Links

<*> To visit your group on the web, go to:
http://groups.yahoo.com/group/commvault/

<*> Your email settings:
Individual Email | Traditional

<*> To change settings online go to:
http://groups.yahoo.com/group/commvault/join
(Yahoo! ID required)

<*> To change settings via email:
commvault-digest < at > yahoogroups.com
commvault-fullfeatured < at > yahoogroups.com

<*> To unsubscribe from this group, send an email to:
commvault-unsubscribe < at > yahoogroups.com

<*> Your use of Yahoo! Groups is subject to:
http://docs.yahoo.com/info/terms/