Avamar: Using volume parallelism to improve Avamar backup performance
Summary: This article discusses the use of volume parallelism to improve Avamar backup performance. This is ONLY for clients that have multiple volumes on distinct, physical devices. An explanation of the plug-in-based strategy is included in the discussion. ...
This article applies to
This article does not apply to
This article is not tied to any specific product.
Not all product versions are identified in this article.
Instructions
Engage support for performance review before implementing the procedures described for filesystem backups.
By default, Avamar scans files in a dataset and handles one partition or volume at a time.
For clients where these multiple volumes are carved from the same disk, this is an efficient approach. It avoids issuing too many simultaneous requests to the read head.
Where Avamar backs up multiple volumes that are located on separate hard drives, this is inefficient. This is because disks that are not being backed up, are idle.
Configuring Avamar to back up volumes in parallel:
Check if the backup covers multiple volumes. In the backup log we might see something like this:
On Windows this can be done using the Disk Management interface in Computer Manager.
The following screenshot shows where it would NOT be appropriate to process volumes in parallel. Note how partitions C: and E: are located on the same hard drive (Disk0).
On Linux clients, use the "mount | grep dev" command:
Here we see that vol01, vol02, and vol03 are on distinct physical devices.
Before applying parallelism, ensure that the volumes are not on the same disk or sharing the same hardware. Enabling the feature in this case is likely to cause disk thrashing and worse performance than the default setting.
For clients where partitions or volumes are located on distinct disks, add the following flags to a new line in the client's avtar.cmd file:
Below are sample logs showing the difference observed on a v7.0.102-47 client before and after the backup ran with volume parallelism.
Without flags:
By default, Avamar scans files in a dataset and handles one partition or volume at a time.
For clients where these multiple volumes are carved from the same disk, this is an efficient approach. It avoids issuing too many simultaneous requests to the read head.
Where Avamar backs up multiple volumes that are located on separate hard drives, this is inefficient. This is because disks that are not being backed up, are idle.
Configuring Avamar to back up volumes in parallel:
Check if the backup covers multiple volumes. In the backup log we might see something like this:
<targetlist>
<path backup="true" name="C:/" />
<path backup="true" name="H:/" />
<path backup="true" name="R:/" />
<path backup="true" name="T:/" />
</targetlist>
Check that these volumes are located on physically distinct devices.
On Windows this can be done using the Disk Management interface in Computer Manager.
The following screenshot shows where it would NOT be appropriate to process volumes in parallel. Note how partitions C: and E: are located on the same hard drive (Disk0).
On Linux clients, use the "mount | grep dev" command:
admin@linuxclient:~/>: mount | grep dev
/dev/sda5 on / type ext3 (rw,acl,user_xattr)
udev on /dev type tmpfs (rw,mode=0755)
tmpfs on /dev/shm type tmpfs (rw,mode=1777)
devpts on /dev/pts type devpts (rw,mode=0620,gid=5)
/dev/sda1 on /boot type ext3 (rw,acl,user_xattr)
/dev/sda3 on /vol01 type xfs (rw,noatime)
/dev/sda7 on /var type ext3 (rw,acl,user_xattr)
/dev/sdb1 on /vol02 type xfs (rw,noatime)
/dev/sdc1 on /vol03 type xfs (rw,noatime)
Here we see that vol01, vol02, and vol03 are on distinct physical devices.
Before applying parallelism, ensure that the volumes are not on the same disk or sharing the same hardware. Enabling the feature in this case is likely to cause disk thrashing and worse performance than the default setting.
For clients where partitions or volumes are located on distinct disks, add the following flags to a new line in the client's avtar.cmd file:
--parallel --parallel-type=by-volume
Below are sample logs showing the difference observed on a v7.0.102-47 client before and after the backup ran with volume parallelism.
Without flags:
2015-10-23 00:52:35 avtar Info <5156>: Backup #145 timestamp 2015-10-23 00:52:35, 683,925 files, 80,274 folders, 431.7 GB (4,649 files, 90.89 MB, 0.02% new)
2015-10-23 00:52:35 avtar Info <6083>: Backed-up 431.7 GB in 32.88 minutes: 788 GB/hour (1,247,862 files/hour)
With flags:
2016-02-11 03:43:07 avtar Info <5156>: Backup #2287 timestamp 2016-02-11 03:43:07, 692,828 files, 82,427 folders, 401.8 GB (986 files, 105.6 MB, 0.03% new)
2016-02-11 03:43:07 avtar Info <6083>: Backed-up 401.8 GB in 14.51 minutes: 1,661 GB/hour (2,864,872 files/hour)
Additional Information
For more information about backup performance optimization, see Avamar: Troubleshooting slow backup performance
For information about adding flags to avtar.cmd, see Avamar: How to gather logs to troubleshoot backup and restore issues
For information about adding flags to avtar.cmd, see Avamar: How to gather logs to troubleshoot backup and restore issues
Affected Products
AvamarProducts
Avamar, Avamar ClientArticle Properties
Article Number: 000156635
Article Type: How To
Last Modified: 12 Sep 2024
Version: 9
Find answers to your questions from other Dell users
Support Services
Check if your device is covered by Support Services.