Professional Documents
Culture Documents
Release 6.5
Publication release 6.5, revision 1
Veritas NetBackup PureDisk Best Practices Guide
Copyright © 2008 Symantec Corporation. All rights reserved.
PureDisk 6.5.0
Symantec, the Symantec logo, and PureDisk are trademarks or registered trademarks of
Symantec Corporation or its affiliates in the U.S. and other countries. Other names may be
trademarks of their respective owners.
The product described in this document is distributed under licenses restricting its use,
copying, distribution, and decompilation/reverse engineering. No part of this document
may be reproduced in any form by any means without prior written authorization of
Symantec Corporation and its licensors, if any.
Symantec Corporation
20330 Stevens Creek Blvd.
Cupertino, CA 95014
www.symantec.com
Technical support
For technical assistance, visit
http://www.symantec.com/business/support/index.jsp. Type
NetBackup PureDisk in the Find Your Product field.
Contents
Chapter 1 Overview
About this manual ................................................................................................. 7
Policy tuning
If you have more than 200 clients, make sure that there are no more than 200
data selections included in one backup policy.
To exclude files from data selections after they have been backed up
1 Recreate the data selection.
2 Disable the schedules on the old data selection.
3 Delete the old data selection after the retention period that you defined in
the data selection removal policy.
these traditional plans write both the file content and the metadata to backup
media, they usually consume large amounts of tape or disk media.
If you have several similar clients, you can create a traditional backup plan for
these clients that employs PureDisk and its data deduplication technology.
Unlike typical backup methods that consume large amounts of backup media,
PureDisk uses its data deduplication technology to store file content only one
time. When you run PureDisk backups at daily, weekly, and monthly intervals,
you ensure that you have a record of the metadata at these intervals. In addition,
PureDisk also backs up any file content that is new or changed since a previous
backup was performed.
To implement this plan, create data selections for each of these intervals, apply
them to clients, and create backup policies. To ensure that these backups are
removed from the system after a suitable retention period, also create three data
removal policies.
The procedures in the following sections describe how to create the policies that
simulate this traditional approach to backing up clients:
■ “Creating data selection templates for daily, weekly, and monthly backups”
on page 11
■ “Applying data selection templates to the clients” on page 12
■ “Creating policies for daily, weekly, and monthly backups” on page 13
■ “Creating data removal policies for daily, weekly, and monthly backups” on
page 15
■ “Creating policy escalation actions” on page 18
To copy a daily-use template and modify it for weekly and monthly use
1 Click the Data Selection Templates tab.
2 In the middle pane, select the daily template that you created in the
previous procedure.
3 In the left pane, click Copy Templates.
4 In the middle pane, click the name of the new copy.
5 In the left pane, click Edit Templates.
6 In the dialog box, change the name of the template.
For example, change the template name to Weekly - My Documents backup
or Weekly - data backup.
Do not change either the inclusion rules or the exclusion rules.
7 Click Save.
8 Perform the preceding steps again to create a monthly template.
For example, change the template name to Monthly - My Documents
backup.
To apply a template
1 Click the Data Management tab.
2 Expand the storage pool in the middle pane.
Expand the tree until you see the client or the group of clients to which you
want to apply the template. For example, you can apply the template to all
the clients in a department or to all the clients in a storage pool.
3 Click the client name, the department name, or the storage pool name.
If you want to apply the template to more than one client, click a
department name or the storage pool name.
Planning data selections and backups 13
Simulating a traditional backup plan
Tip: You may have several clients in a department, and you may have
applied this template to some of the clients in a previous session. If you
apply the template on the department level, it is faster than to apply the
template on each client individually. PureDisk does not reapply the
template on the older clients. However, when you apply the template on the
department level, you do not need to select each new client individually and
apply the template.
To copy a daily use backup policy and modify it for weekly use
1 Click the Workflows tab.
2 In the middle pane, click the daily policy that you created in the previous
procedure.
3 In the left pane, click Copy Policy.
4 In the middle pane, click the name of the new copy.
5 In the left pane, click Edit Policy.
6 Edit the tabs in the dialog box.
Perform the following steps:
■ On the General tab, change the policy name. For example, change the
policy name to Weekly backup.
■ On the Data Selections tab, use the Data selections based on template
drop-down list to specify the name of the weekly backup data selection.
See “Creating data selection templates for daily, weekly, and monthly
backups” on page 11.
■ On the Scheduling tab, perform the following steps:
■ In the upper part of the tab, specify a start time and select Weekly
schedule.
■ In the lower part of the tab, click in the boxes to select Saturday.
■ On the Parameters tab, select parameters that are appropriate to this
data selection for your site.
7 Click Save.
Planning data selections and backups 15
Simulating a traditional backup plan
To copy a weekly use backup policy and modify it for monthly use
1 Click the Workflows tab.
2 In the middle pane, click the weekly policy that you created in the previous
procedure.
3 In the left pane, click Copy Policy.
4 In the middle pane, click the name of the new copy.
5 In the left pane, click Edit Policy.
6 Edit the tabs in the dialog box.
Perform the following steps:
■ On the General tab, change the policy name. For example, change the
policy name to Monthly backup.
■ On the Data Selections tab, use the Data selections based on template
drop-down list to specify the name of the weekly backup data selection.
See “Creating data selection templates for daily, weekly, and monthly
backups” on page 11.
■ On the Scheduling tab, perform the following steps:
■ In the upper part of the tab, specify a start time and select Monthly
schedule.
■ In the lower part of the tab, click in all boxes to select all months in
the year.
■ On the Parameters tab, select parameters that are appropriate to this
data selection for your site.
7 Click Save.
2 In the middle pane, click Data Management Polices > Data Removal.
3 In the tasks pane, click Create Policy.
4 Complete all the tabs in the policy creation dialog box.
Perform the following steps:
■ On the General tab, perform the following steps:
■ Specify a name for the policy. For example, Daily backup - removal
policy.
■ Select Enabled.
■ On the Data Selections tab, perform the following steps:
■ Expand the tree until you see the client or the group of clients to
which you want to apply the policy.
■ Select the client or the client group. For example, you can apply
the policy to all the clients in a department or to all clients in a
storage pool.
■ On the Data Selections tab, use the Data selections based on
template drop-down list to specify the name of the weekly backup
data selection.
See “Creating data selection templates for daily, weekly, and
monthly backups” on page 11.
■ On the Scheduling tab, perform the following steps:
■ In the upper part of the tab, specify a start time and select Weekly
schedule.
■ In the lower part of the tab, click in the boxes to select Monday,
Tuesday, Wednesday, Thursday, and Friday.
■ (Conditional) On the Metadata tab, you can specify to remove only
specific files.
■ On the Parameters tab, select parameters that are appropriate to this
data selection for your site.
In Remove versions backed up, select Older than (in days) and specify
14 days.
5 Click Save.
To copy a daily data removal policy and modify it for weekly use
1 Click the Workflows tab.
2 In the middle pane, click the daily data removal policy that you created in
the previous procedure.
3 In the tasks pane, click Copy Policy.
Planning data selections and backups 17
Simulating a traditional backup plan
To copy a weekly data removal policy and modify it for monthly use
1 Click the Workflows tab.
2 In the middle pane, click the weekly data removal policy that you created in
the previous procedure.
3 In the left pane, click Copy Policy.
4 In the tree pane, click the name of the new copy.
5 In the left pane, click Edit Policy.
6 Complete all the tabs in the dialog box.
Perform the following steps:
18 Planning data selections and backups
Simulating a traditional backup plan
■ Select Enabled.
■ On the Data Selections tab, select Include all data selections selected
above.
■ On the Scheduling tab, specify a schedule.
■ (Conditional) On the Metadata tab, click Add. Fill in the following fields
in the dialog box to describe the files you want to remove:
■ For Rule name, specify a name for this filter. For example,
MP3 Files.
■ For Folder name, specify an asterisk (*).
■ For File name, specify *.mp3.
5 Click OK.
6 Click Save.
Chapter 4
Comprehensive policy
scheduling
This chapter includes the following topics:
■ “Scheduling backup, replication, and maintenance policies” on page 23
Backup policies No default Run backup policies on work days Schedule the backup policies to run when
when system activity is high. system activity is low, for example, at
night between 8pm and 8am.
For example, configure the backup
window in the Parameters tab for 8pm
through 8am.
24 Comprehensive policy scheduling
Scheduling backup, replication, and maintenance policies
Table 4-1 PureDisk recommended practices for policies and scripts (Continued)
Data Removal No default Run a removal policy one, two, or Configure this policy right after you
policy three times a week and when few deploy a storage pool. Do not wait for
backup policies run. content routers to fill with unneeded
backups.
The more frequently you run a data
removal policy, the less time each run
takes because less data is removed in each
run.
If you cannot schedule frequent data
removal policy runs, run the data
removal policy one time a week, on the
weekend. If you schedule this policy to
run less frequently, you risk overloading
the system when it runs.
CR Garbage Monthly Monthly. Run the policy at least one time a month.
Collection
policy
Data Selection No default Run as needed. Run the policy when you delete a whole
Removal policy data selection on a client.
Table 4-1 PureDisk recommended practices for policies and scripts (Continued)
MB Garbage Monthly One to three times each week. Check Log deleted records if you want to
Collection create a log of deleted records in the job
policy log of the metabase garbage collection
policy.
If you check Enable extensive cleanup, be
aware that this method consumes many
system resources. Enable this option on
an infrequent basis, for example one time
every three months. You can copy the
default policy and enable this capability
only in a policy that runs every three
months.
Content router No default Four times each day. Performs content router queue
queue maintenance.
Schedule this policy to run outside
processing your file backup and disaster Do not disable this policy. Make sure this
policy
recovery backup windows. policy runs regularly.
26 Comprehensive policy scheduling
Scheduling backup, replication, and maintenance policies
Table 4-1 PureDisk recommended practices for policies and scripts (Continued)
Configuration parameters
FingerprintType
This parameter determines the fingerprinting algorithm to be used. Type 0 is
the old algorithm without fingerprint collision detection and type 1 is for
PureDisk release 6.5 with fingerprint collision detection. This value is set during
agent installation and should not be changed.
Port
This parameter determines the port on which the content routers listen for
incoming connections. If you change this parameter, you must enter the same
number in the configuration file of all content routers and all agents. Symantec
recommends that you leave this parameter unchanged.
TCPKeepAlive
This parameter determines whether or not TCP keep-alive probes are to be sent
during connection idle time. It allows an agent to detect if an existing
connection with a content router is still valid.
For more information regarding this topic, search Google on the Web for
SO_KEEPALIVE and TCP_KEEPALIVE or visit the Web pages listed in the
TCPSendBufferSize parameter.
28 Tuning PureDisk
Configuration parameters
TCPSendBufferSize
This parameter sets the maximum socket send buffer size in bytes. If you
increase this value, the agent can transmit more data before the backup blocks.
The backup blocks while it waits for more buffer space to become available,
which happens after PureDisk sends the currently buffered data to the content
router. The default value is operating-system dependent and for UNIX, usually
depends on other kernel configuration settings. Search Google for SO_SNDBUF
for more information on this subject.
On high-latency lines (or high bandwidth lines), it is definitely worthwhile to
experiment with various values for the send buffer size. For optimal behavior,
the TCPReceiveBufferSize buffer size should be adjusted as well.
A general formula for establishing an initial value for this parameter in
kilobytes is as follows:
size = b * d * 10^3 / 8
The formula assumes a delay d in microseconds and a bandwidth b in
Kbytes/sec. You can measure the delay in microseconds by using the ping
command)
If you have a 45M-bit line and a 30-ms delay, the optimal send buffer size would
be 165Kbytes.
http://dast.nlanr.net/Guides/GettingStarted/TCP_window_size.
html has additional information on this subject, and
http://www.die.net/doc/linux/man/man7/tcp.7.html describes the
various kernel configuration parameters in detail.
If you specify a value of 0, it enables the operating system default.
TCPReceiveBufferSize
This parameter sets the maximum socket receive buffer size, in bytes. If you
increase this value, it may reduce the number of disk writes that the agent must
perform during a restore. Using a small value has a negative impact on
performance. The default value is operating-system dependent and for UNIX,
usually depends on other kernel configuration settings. Search Google for
SO_RCVBUF for more information on this subject.
If you specify a value of 0, it enables the operating system default.
MaxTransferRate
This parameter specifies, in Kbytes/sec, the maximum throughput on a
connection with a content router (with a variance of about 10%). A value of 0
means no bandwidth limit is applied.
Tuning PureDisk 29
Configuration parameters
ReadBufferSize
This parameter specifies, in bytes, the maximum amount of data that can be
read from disk in a single read operation. The default value is 64kbytes, which
corresponds to the default TCPSendBufferSize on most operating systems.
If a backup consists mainly of large files a higher value could improve
performance slightly. If you specify a lower value, it has a negative impact on
performance because PureDisk requires more read operations to process a file.
An optimal value is not easy to establish; it requires experimentation with
different values. If you specify large values, it may have an adverse effect on
performance because it may defeat the read-ahead capabilities of the operating
system or underlying hardware.
Initially, the following values may be used:
For an average file size < 128KB use the default setting
For an average file size < 64MB use 128kb
For an average file size < 128MB use 256kb
For an average file size> 128MB use 512kb
No direct link exists between this configuration parameter and the
TCPSendBufferSize parameter. However, there is an indirect connection
between the two parameters. The TCP buffer configuration directives let you
specify settings to result in optimal network throughput. The Read (and Write)
buffer directives let you specify settings to achieve optimal disk throughput.
There also exists an indirect link with the specified segment size: the agent
needs to buffer data until it has a full segment available so that it can compute
the segment fingerprint. To some degree, the rules that are given here use this
link.
Note: If you specify a higher value, more memory is used during backup.
WriteBufferSize
This parameter specifies, in bytes, the maximum amount of the data the agent
may buffer during a restore before writing it to disk. The default value is 32k
30 Tuning PureDisk
Configuration parameters
B
backup
implementing a traditional backup plan 10
initial 9
policy recommendations 23
PureDisk timeouts 9
snapshot support 10
best practices 9, 21, 23, 27
C
content router queue processing
recommendations 25
D
data deduplication 11
data reduction factor, see data deduplication
data removal policy
for specific file types 21
recommendations 24
data selection
excluding files after a backup 10
removal policy recommendations 24
disaster recovery
policy recommendations 24
G
garbage collection
policy recommendations 25
M
maintenance
policy recommendations 25
R
replication
policy recommendations 24
Index 32