Faster and more space efficient backup for Microsoft SQL Server

Faster and more space efficient backup for Microsoft SQL Server
Faster and more space efficient backup
for Microsoft SQL Server databases
HPE StoreOnce Plug-in for Microsoft SQL Server
Contents
Executive summary.........................................................................................................................................................................................................................................................................2
Introduction to Microsoft SQL Server backup and restore .........................................................................................................................................................................3
HPE StoreOnce Backup Systems—Key features and benefits ...............................................................................................................................................................4
Introducing HPE StoreOnce Catalyst .......................................................................................................................................................................................................................5
HPE StoreOnce Plug-in for Microsoft SQL Server ..............................................................................................................................................................................................5
Configuration guidance.........................................................................................................................................................................................................................................................6
Creating multiple backup copies ..................................................................................................................................................................................................................................8
Using Federated Catalyst stores...................................................................................................................................................................................................................................8
Using Fibre Channel connection to Catalyst stores ....................................................................................................................................................................................9
Password-protected access to Catalyst stores for added security ......................................................................................................................................................9
HPE StoreOnce Plug-in for Microsoft SQL Server in action ..................................................................................................................................................................11
Backup with MIRROR TO clause—Monitoring performance .........................................................................................................................................................12
Automating daily tasks ......................................................................................................................................................................................................................................................15
Viewing backups on the Catalyst store ..............................................................................................................................................................................................................16
Point-in-time restore ...........................................................................................................................................................................................................................................................17
Backup with Asynchronous Copy—Monitoring performance .......................................................................................................................................................18
Backup with Copy Redirect—Monitoring performance.......................................................................................................................................................................22
Performance ......................................................................................................................................................................................................................................................................................26
Known limitations..........................................................................................................................................................................................................................................................................29
Summary ...............................................................................................................................................................................................................................................................................................30
Appendix A—HPE StoreOnce Catalyst overview............................................................................................................................................................................................30
Supported device types ....................................................................................................................................................................................................................................................30
StoreOnce Catalyst terminology and summary features ...................................................................................................................................................................30
Technical white paper
Technical white paper
Page 2
Executive summary
Organizations worldwide run Microsoft® SQL Server databases to support their mission-critical applications. The two main concerns from a backup
perspective in SQL Server database environments have high data growth rates and shrinking backup windows. HPE StoreOnce Backup appliances
provide highly efficient deduplicating storage that is accessed using virtual tape, NAS, and HPE StoreOnce Catalyst device types. HPE StoreOnce
Catalyst is purpose-designed for backup and recovery workflows, without the overheads and limitations of general purpose protocols, and as a result
is the highest performing and most configurable of these device types. Using a Catalyst-capable backup application software or an HPE StoreOnce
Plug-in, the reduction of duplicate backup data can be executed at the database server (source-side deduplication) to reduce network bandwidth
requirements for backup and enable faster backups. For added data protection, deduplication-optimized replication of backups can also be configured.
The HPE StoreOnce Plug-in for Microsoft SQL Server enables direct backup of SQL Server databases to Catalyst stores with faster throughput
and greater deduplication ratios than other backup solutions, thus helping to solve the issues associated with high data growth rates and
shrinking backup windows for Microsoft SQL Server databases.
In many organizations, a key role of the MS SQL Server Database Administrators (DBAs) is to manage the availability of the databases. A DBA who is
responsible for database availability should be able to rely on backups as the last resort, during any disaster event, to restore lost or corrupt data.
Failing to have reliable backups can cost the DBA his or her job. To avoid this situation, the DBA should have total visibility into the control of backup
and disaster recovery (DR) processes without involving the backup administrator. The StoreOnce Plug-in for MS SQL empowers DBAs to manage
the entire backup, recovery, and DR process using standard MS SQL BACKUP and RESTORE T-SQL commands. Today, MS SQL DBAs generally use
local disks or CIFS shares as the target device for backups. This paper describes the many advantages of using HPE StoreOnce Catalyst as the target
device for MS SQL Server backups. In summary, these benefits are:
• Improved backup throughput by means of:
– Catalyst protocol, which is more efficient than the CIFS protocol
– Distributing the deduplication load across several MS SQL Servers rather than having the single backup target doing all the deduplication
processing
– Automatic selection of the optimal number of output streams, based on the number of data files in the database
• Improved storage savings by means of the HPE intelligent processing included in the plug-in.
• Use of syntaxes for BACKUP/RESTORE DATABASE/LOG commands very similar to their T-SQL equivalent, placing the DBA in his or her
comfort zone.
• Up to three additional copies of the backup can be made using MS SQL MIRROR TO directive and sent to DR sites.
• Supported on 64-bit (x64) versions of MS SQL Server 2008, 2012, and 2014, under 64-bit (x64) versions of Windows Server® 2008 and
2012.
• The HPE StoreOnce Plug-in for Microsoft SQL Server is free of charge and downloadable from hpe.com/go/softwaredepot.
Usage notes
• When using the StoreOnce Plug-in for MS SQL Server, note that consideration needs to be given to the additional CPU requirements on the
MS SQL Server to support the StoreOnce Catalyst Plug-in, and if configured, source-side deduplication. When using source-side deduplication,
a large part of the deduplication is done on the SQL Server itself.
• An HPE StoreOnce Catalyst license must be purchased and installed on the StoreOnce appliance hosting the target Catalyst store in order for
the plug-in to write to the target appliance.
The HPE StoreOnce Plug-in for Microsoft SQL Server concept is shown in Figure 1.
Technical white paper
Page 3
MS SQL Server Administrator
MS SQL database
StoreOnce Plug-in for
MS SQL Server
Parallel streams
Intelligent
data
processing
MS SQL Server
Low- or
highbandwidth
backup
HPE StoreOnce
Catalyst store
Figure 1. HPE StoreOnce Plug-in concept.
Target audience: This white paper is intended to assist MS SQL Server Administrators who are responsible for backups as part of their responsibility
for database availability. Generally, they prefer to use the standard MS SQL Server tools such as SQLCMD or SQL Maintenance Tasks to back up
directly to disk or NAS share and maintain total control of the entire MS SQL Server environment. However, this may not make optimal use of the
backup storage or backup window available.
Using the HPE StoreOnce Plug-in for Microsoft SQL Server, the DBA can achieve improved backup performance, using lower bandwidth with
higher deduplication ratios to save on network and storage costs and to get the backups finished faster.
This white paper describes how this technology works and how to take advantage of it.
Assumptions: The bulk of this white paper assumes a working knowledge of Microsoft SQL database structures, BACKUP/RESTORE
DATABASE/LOG T-SQL syntaxes, and HPE StoreOnce Catalyst. For those not familiar with MS SQL Server BACKUP/RESTORE syntaxes, the
user guide lists the available options.
Introduction to Microsoft SQL Server backup and restore
Microsoft SQL Server contains all the components needed to run backup and restore of databases and Transaction logs. By using SQL Maintenance
Tasks or scripting SQLCMD, DBAs can back up to local disks or CIFS shares. MS SQL Server also contains an application program interface (API) that
is able to integrate backup and restore operations into third-party backup software such as HPE Data Protector or Symantec NetBackup. This
interface is called Virtual Backup Device Interface and comes built-in with all recent versions of MS SQL Server.
The HPE StoreOnce Plug-in for MS SQL Server is integrated with MS SQL Server the same way as a third-party software. It uses VDI to submit
backup and restore commands to MS SQL Server. It then handles the stream and writes or reads it to or from a Catalyst store.
The newest version of the HPE StoreOnce Plug-in for MS SQL Server v1.2 incorporates a GUI that allows a DBA to handle all functions from
creating backup policies, restoring a backup, and viewing items stored on Catalyst stores. Throughout this paper, screenshots are included with
accompanying CLI commands. Figure 2 below shows the creation of a backup policy, which would do the exact same as the CLI command
“BACKUP DATABASE model TO STORE=“StoreOnce1/SQL” WITH MAXTRANSFERSIZE=4194304, CHECKSUM”.
The command the DBA specifies to start a backup to a Catalyst store is very similar to the T-SQL command the DBA would use to run a backup
from SQL Management Studio (or SQLCMD) to a CIFS share. For example:
Technical white paper
Page 4
Backup to a CIFS share:
BACKUP DATABASE model TO DISK=“\\StoreOnce1\SQL\model_2016-08-01_134532.bak” WITH MAXTRANSFERSIZE=4194304,
CHECKSUM
Backup to a Catalyst store using the plug-in:
BACKUP DATABASE model TO STORE=“StoreOnce1/SQL” WITH MAXTRANSFERSIZE=4194304, CHECKSUM
The main differences between the two solutions are
• The TO DISK= becomes a TO STORE=.
• You do not specify the backup name when using the plug-in. A unique name containing the server name, instance name, DB name, and
timestamp is generated.
Figure 2. HPE StoreOnce Plug-in policy creation for a backup policy to a Catalyst store.
HPE StoreOnce Backup Systems—Key features and benefits
HPE StoreOnce deduplication—Store more data on disk
HPE StoreOnce deduplication reduces the disk space required to store backup data sets without impacting backup performance. Retaining more
backup data on disk longer, enables greater data accessibility for rapid restore of lost or corrupt data and reduces downtime. Deduplication ratios
are strongly influenced by two factors—data change rate and backup data retention periods. Low data change rates and data retained for longer
periods of time yield higher deduplication ratios. Figure 3 shows how much more data can be stored on StoreOnce.
Deduplication-enabled replication
HPE StoreOnce deduplication is the technology enabler for HPE StoreOnce deduplication-enabled replication, which allows fully automated
replication over low-bandwidth links to a DR site, giving remote office/branch office (ROBO) and small data centers a cost-effective DR solution.
Deduplication-enabled replication is fully supported in the plug-in when the backup is run with one or more MIRROR TO STORE= clause.
Technical white paper
Page 5
Rapid restore of data for dependable, worry-free data protection
HPE StoreOnce appliances offer immediate access to backups for rapid restores. HPE StoreOnce deduplication allows more data to be stored
closer to the data center for longer periods of time, which offers immediate access for rapid restores. A lot of design considerations have gone
into achieving high restore performance from the HPE StoreOnce Backup Systems. HPE Data Assurance technology—adding checksums at key
stages of the deduplication process enables very high levels of data integrity.
Simplify and improve the backup process
HPE StoreOnce appliances can back up multiple servers via a standard Ethernet or Fibre Channel (FC) network simultaneously to a disk-based
solution at peak speeds of up to 137 TB/hr, for the largest appliance, meaning that substantially reduced backup windows are possible.
Terabytes
HPE StoreOnce appliances can be intuitively managed and configured using the browser-based administrative interface. HPE StoreOnce Systems
are largely self-managing backup appliances that require little routine maintenance. Unlike other disk-based storage devices, HPE StoreOnce
Systems do not require virus protection or LUN provisioning.
Usage model 1
Usage model 2
Usage model 3
1% rate of change
Daily full
5% rate of change
Daily full
5% rate of change
Weekly full/daily incr.
Total backup data
Size on HPE StoreOnce
Space saved (deduplication)
Figure 3. HPE StoreOnce deduplicated data stored on StoreOnce vs. the size of the backup.
Introducing HPE StoreOnce Catalyst
HPE StoreOnce Catalyst is a purpose-designed backup and recovery device type. It can be accessed over Ethernet or FC. StoreOnce Catalyst
targets are recommended for large and complex data-protection environments where highest performance and flexibility of data movement is
required. StoreOnce Catalyst can increase backup speeds and reduce backup load on the network by configuring distributed deduplication
processing to enable operation in low-bandwidth mode. The same deduplication-optimized data transmission enables highly efficient movement
of backup data between StoreOnce appliances for added resiliency to potential data loss incidents. For more details, see Appendix A.
HPE StoreOnce Plug-in for Microsoft SQL Server
HPE StoreOnce Plug-in for Microsoft SQL Server delivers control of MS SQL backup and disaster recovery, enabling DBAs to have confidence in
self-administered recovery from the local or the DR site—without dependence on a backup administrator. In addition, the HPE StoreOnce Plug-in
for Microsoft SQL Server provides accelerated backup performance over traditional CIFS disk targets. When used across large number of MS SQL
application servers each performing source-side deduplication, the combined effect of distributing the deduplication processing enables faster
backup throughput, provided the application servers themselves are sized correctly.
Technical white paper
Page 6
Figure 4 shows the two modes of operations that underpin this technology when the plug-in is installed on the MS SQL Server. Referred to as
target-side deduplication (high-bandwidth backup) or source-side deduplication (low-bandwidth backup), both modes feature intelligent MS
SQL Server data processing where the format of the MS SQL Server backup stream is understood and optimized by the plug-in to increase space
saving and backup speed. In target-side deduplication, all the deduplication processing takes place on the HPE StoreOnce appliance with all the
backup data transferred across the network. In source-side deduplication (low-bandwidth mode), the majority of the deduplication process is
performed on the MS SQL Server itself and only unique chunks of backup data are sent over the network to the HPE StoreOnce appliance, with
significant savings on network bandwidth.
Hewlett Packard Enterprise recommends using the low-bandwidth mode on Catalyst stores used with the HPE StoreOnce Plug-in for MS SQL
Server. Using low-bandwidth mode deliver significantly faster backup and improved space savings for many use cases.
Target-side deduplication (high-bandwidth backup)
Intelliigent
data
processing
Write
Catalyst client API
Catalyst store
Send all data
Buffer
Chunk
Hash
Match
Compress
Store
Source-side deduplication (low-bandwidth backup)
Send hashes
Intelliigent
data
processing
Catalyst store
Catalyst client API
Write
Chunk
Hash
Match
Receive matches
Store
Compress
Send unique chunks
Figure 4. HPE StoreOnce Plug-in for MS SQL Server—How it works.
The HPE StoreOnce Plug-in for Microsoft SQL Server v1.2 is supported on
• Database servers: x64 versions of SQL Server 2008, 2008 R2, 2012, 2014
• Operating systems: x64 versions of Windows Server 2008 SP2, 2008 R2, 2012, 2012 R2
For backup to Catalyst device types:
• All StoreOnce appliances with StoreOnce 3.9 or newer software
For the latest compatibility information, please visit hpe.com/go/ebs.
Configuration guidance
The HPE StoreOnce Plug-in for Microsoft SQL Server is optimized to back up large databases using multiple data files located on different
partitions or mount points, as recommended by the Microsoft best practices for mission-critical databases.
The plug-in automatically splits the backup stream generated by the SQL Server into several Catalyst items, all copied in parallel to the StoreOnce
appliance. By default, the plug-in creates N+2 output items in the Catalyst store, where N is the number of data files in the database. Unless the
backup targets a StoreOnce appliance with a maximum concurrent stream count less than N+2, it is recommended to use the default value. Should
you need to decrease the value, for example, to back up a StoreOnce Virtual Storage Appliance (VSA), the parameter –maxstreams=n can be
specified. Note, however, that a lower value does impact negatively the backup performance and deduplication ratio.
To increase backup throughput performance and storage space savings, the following guidelines should be followed.
Technical white paper
Page 7
• Set the Catalyst store with Primary (default) transfer policy: Low bandwidth (It is the default option when creating new Catalyst stores.).
This mode enables the deduplication load to be split between the database server and the StoreOnce appliance. The amount of data to be
transferred between the SQL Server and the StoreOnce Application is consequently decreased significantly. The low-bandwidth mode also
makes possible internal optimization for better performance not available with the high-bandwidth mode.
• Run the SQL backup with a large value of the SQL parameter maximum transfer size, for example, “WITH MAXTRANSFERSIZE=4194304.”
• Do not specify any value for the parameter –maxstreams. Leave the plug-in to compute a default value.
Figure 5. Effect of multiplexing on throughput using a SQL Server database with eight data files.
Note
The plug-in does not support more than one output device as it already splits the stream into several Catalyst items. Therefore, do not attempt
to specify multiple output devices to the SQL BACKUP command.
The first backup of a database is generally slower than subsequent backups. This is because it is likely that all or most of the chunks that make up
the first backup do not exist in the Catalyst store so have to be compressed and transmitted to the StoreOnce appliance. As more backups are
written to the Catalyst store, it is likely that more duplicate chunks can be identified so fewer unique chunks have to be compressed and transmitted
making the backups faster. It is the same when backup copies are configured. The first time a copy is made it is likely that the copy requires more
data to be transmitted and takes significantly more time than subsequent copies.
Encryption and compression
As usual, when sending backups to deduplication capable devices, it is recommended not to compress or encrypt backups upstream of deduplication,
as this can generate unique data for every backup, thus significantly reducing the backup throughput and deduplication ratio. StoreOnce does
compress backups, to reduce space usage, but it does so after unique chunks have been identified so deduplication efficiency is not reduced. For
database backups that need to be encrypted, the StoreOnce Security Pack option enables on disk encryption and data-in-flight encryption.
Using scripts to automate backups
To automate backups, scripts need to be written using the StoreOnce Plug-in for SQL Server commands, typically using Microsoft PowerShell. To
initiate the scripts, the Windows® Scheduler is used.
Technical white paper
Page 8
Creating multiple backup copies
The HPE StoreOnce Plug-in for Microsoft SQL Server also supports the MIRROR TO SQL option to create up to four copies of the backup
(including the primary or local copy). These copies can be copied over a WAN to remote sites because in low-bandwidth mode, only unique data
is sent across the network. An initial seeding backup or copy to all locations is required, but this is a one-off occurrence during the first backup.
When a TO STORE=“StoreOnce1/Store1”, MIRROR TO STORE=“StoreOnce2/Store2” is specified, the plug-in first executes a backup to
StoreOnce1/Store1, and then instructs the StoreOnce1 to make a copy of this backup to StoreOnce2/Store2. The program exits only when all
copies are done.
Up to three MIRROR TO STORE can be specified, for a total of four copies of a backup. When creating a new policy in the GUI, the DBA can
specify to use the MIRROR TO STORE option, and set up the backup and copy jobs, just like the CLI command.
Figure 6. Policy creation that utilizes the MIRROR TO SQL option.
Using Federated Catalyst stores
Federated Catalyst stores can be created on StoreOnce multinode appliances. A Federated Catalyst store is of value when large databases are
being protected by backup to a multinode StoreOnce appliance, and for ease of management, a larger backup target is possible with a single
Catalyst store. The recommendation for regular Catalyst store is not to grow the store bigger than the capacity of a single node. A Federated
Catalyst store can logically combine capacity from up to eight nodes to create a much bigger backup device. To the DBA, the best practices for
using the device and the resulting performance characteristics of a Federated Catalyst store are the same as a regular Catalyst store.
Technical white paper
Page 9
Using Fibre Channel connection to Catalyst stores
The configuration is very similar when backing up over Fibre Channel as when backing up over Ethernet. The difference is that a Catalyst over Fibre
Channel (CoFC) identifier is used, instead of a Catalyst over Ethernet (CoE) identifier to specify the target Catalyst device. The following example
demonstrates a backup of the database model to the Federated Catalyst store TC-SQL, using the CoFC identifier of one of the StoreOnce appliances:
“C:\Program Files\Hewlett-Packard Enterprise\StoreOnce\isvsupport\sql\bin\HPStoreOnceForMSSQL.exe” -backup “BACKUP
DATABASE [model] TO STORE=‘COFC-hp3ca82a1f59f802/TC-SQL’ WITH MAXTRANSFERSIZE=4194304, CHECKSUM” tag=TestingCoFCandFederatedCatalyst
As described previously, the HPE StoreOnce Plug-in for Microsoft SQL Server supports writing backups over Fibre Channel to Catalyst and
Federated Catalyst stores. There is a restriction that Catalyst stores and Federated Catalyst stores accessed over Fibre Channel cannot be the
target of a Catalyst Inbound Copy Jobs. In other words, a Fibre Channel connection can be used in a TO STORE clause, but cannot be used in a
MIRROR TO STORE clause. With the addition of the Copy Redirect feature in the newest release of the plug-in, the copy target can now be
Catalyst over Fibre Channel but uses Catalyst over Ethernet to facilitate the transfer of data between StoreOnce devices.
Figure 7 below shows how the GUI shows the DBA a registered CoFC Store. The identifier is clearly displayed, along with the name of the
Catalyst store. This is almost the same as a CoE Catalyst store, except that the CoFC identifier is used.
Figure 7. Registering a Catalyst Store over Fibre Channel in the plug-in GUI.
Password-protected access to Catalyst stores for added security
To reduce the risk of unauthorized access to the SQL Server backup data, user names and passwords can be configured to control access to the
backup targets (Catalyst stores) and simple scripts used to leverage Windows Credential Manager to securely store the user names and passwords.
The StoreOnce appliance administrator configures the Catalyst store with the access credentials. The SQL Server Administrator can use the sample
scripts as follows to integrate the access credentials with Windows Credential Manager.
Catalyst client access permissions
In many organizations, the administrator of the StoreOnce appliance will not be the DBA. StoreOnce administrator rights are required to create
the Catalyst store that can be used as the backup target for the SQL Server backups. The administrator can set client access permissions at the
time the Catalyst store is created. The client (user name) and password are assigned in the client’s tab of the Catalyst Configuration GUI page
(Figure 8). For use with the StoreOnce Plug-in for SQL Server all the permissions options should remain unchecked. The user name, password,
appliance location, and store name are shared with the SQL Server DBA to enable access to the backup target. If a user name and password is
not set, all Catalyst clients will be able to access the Catalyst store.
Technical white paper
Page 10
Figure 8. The StoreOnce appliance administrator creates a client (SQL_DBA) for the Database Administrator to use and enables access to the backup target (Store_1).
Using Windows Credential Manager to manage user names and passwords
The following PowerShell code samples are an example of how to use Windows Credential Manager to store user names and passwords that the
plug-in requires to access password-protected Catalyst stores. Two scripts are required as in the following example. This example uses a file called
credentials.txt.
Note
Using Windows Credential Manager will protect your Catalyst store access details from all the Windows accounts except those with administrator
privileges.
Sample script 1: Store user name and password in a secure text file
This script is run once, before starting backup operations, per DBA (user name and password pair) to generate a secure file to store the credentials.
Running this script prompts the user to enter the credentials of the store (user name and password) via the Windows Credential Manager dialogue
box. The script converts the user name to a secure string and the user name and password are stored securely in the specified file. It is not necessary
to convert the password to a secure string since the Windows credentials object already stores the password as a secure string. The sample code for
script 1 is in Figure 9.
$FileString = “C:\credentials.txt”
$Credential = Get-Credential
# With the following command, we get the username in a secure string
$username = convertto-securestring $Credential.username -asplaintext -force
$password = $Credential.Password
# To save username and password to file
$username | ConvertFrom-SecureString | Set-Content $FileString
$password | ConvertFrom-SecureString | Add-Content $FileString
Figure 9. Script example to write backup target credentials to a secure file.
Note
If the DBA loses the Windows Credential Manager account, the user name and password are not recoverable. In this case, the script 1 will need to
be rerun.
Technical white paper
Page 11
Sample script 2: Access the user name and password
This is integrated with any script that will access a backup target that is password protected. The script in Figure 9 reads the secure file created by
the script in Figure 10 and converts the secure string back to plain text. The user name and password extracted from the file are passed as parameters in
the backup script. The user name and password are passed as plain text but no other user can to convert the entries in the file to plain text because only
the Windows user who created the file can access the credentials. The process must be executed for each password-protected backup.
$FileString = “C:\credentials.txt”
# Get credentials from file
$credentials = Get-Content $FileString | ConvertTo-SecureString
# Separate username and password (username is line 1 whereas password is stored in line 2)
$username = convertfrom-securestring -securestring $credentials[0]
$password = convertfrom-securestring -securestring $credentials[1]
# Convert username and password which are currently stored as secure string into plain text.
$PlainTextUserName =
[Runtime.InteropServices.Marshal]::PtrToStringAuto([Runtime.InteropServices.Marshal]::SecureStringToBSTR( (Conver
tTo-SecureString $username) ))
$PlainTextPassword =
[Runtime.InteropServices.Marshal]::PtrToStringAuto([Runtime.InteropServices.Marshal]::SecureStringToBSTR( (Conver
tTo-SecureString $password) ))
# This script needs to be in the folder where “HPStoreOnceForMSSQL.exe” is located
\HPStoreOnceForMSSQL.exe -backup “BACKUP DATABASE [master] TO
STORE='$PlainTextUserName`:$PlainTextPassword@123.45.67.89/Store_1' WITH MAXTRANSFERSIZE=4194304,
CHECKSUM”
Figure 10. Sample backup script including access to Catalyst store credentials stored in a secure file.
HPE StoreOnce Plug-in for Microsoft SQL Server in action
This section illustrates how the plug-in can be used with the following example tasks:
• Backup of a database to a local StoreOnce—The backup is then copied to another StoreOnce at the central site.
• Monitoring performance.
• Automating daily tasks.
• A point-in-time restore, using Full, Differential, and Transaction log backups.
• Copy Redirect.
• Asynchronous Copy.
Note
The following examples contain performance information that was achieved with the following configuration:
Host details:
• HPE ProLiant DL380p Gen8 Server with Windows Server 2012 R2 Enterprise edition
• 2.90 GHz CPU
• 64 GB RAM
• QLogic 8 Gb Fibre Channel HBA
• HPE NC552SFP 10GbE 2-port Ethernet Server Adapter
• SQL Server 2014
Technical white paper
Page 12
Storage details:
• HPE 3PAR StoreServ 8400
Switch details:
• HPE SN6000B 16 Gb Fibre Channel Switch
• H3C S12518 10GbE Ethernet Switch
Backup target details:
• HPE StoreOnce 6600 (3.14-1-1607.1)
• HPE StoreOnce 3520 for Copy Redirect Testing (3.15.1-1634.4)
Backup with MIRROR TO clause—Monitoring performance
A Full backup of the database TestDB is sent to two Catalyst stores on two different devices. The first one, at address 10.1.5.21, is a local
StoreOnce. The second device, at address 10.1.5.23, is a secondary node on the StoreOnce at the central site.
Note
This example includes extracts from the plug-in output that shows a very high deduplication ratio. This was achieved using backups where the
database had a very low change rate between backups. Should the database be subject of a large amount of changes, a lower deduplication ratio
and backup throughput may be observed.
The monitoring example uses the StoreOnce appliance GUI. To access this, the DBA requires user access rights to the StoreOnce appliance.
“C:\Program Files\Hewlett-Packard Enterprise\StoreOnce\isvsupport\sql\bin>HPStoreOnceForMSSQL.exe -backup” BACKUP Database
[TestDB] TO STORE='Admin:admin@10.1.5.21/StoreOnce_SQL_catalyst1' MIRROR TO
STORE='Admin:admin@10.1.5.21/StoreOnce_SQL_catalyst2' WITH MAXTRANSFERSIZE=4194304, CHECKSUM,STATS=10
Processed 2492032 pages for database 'TestDB', file 'TestDB' on file 1.
Processed 2491904 pages for database 'TestDB', file 'TestDB2' on file 1.
Processed 2491904 pages for database 'TestDB', file 'TestDB3' on file 1.
Processed 2491904 pages for database 'TestDB', file 'TestDB4' on file 1.
Processed 2491904 pages for database 'TestDB', file 'TestDB5' on file 1.
Processed 2491904 pages for database 'TestDB', file 'TestDB6' on file 1.
Processed 2491904 pages for database 'TestDB', file 'TestDB7' on file 1.
Processed 2492024 pages for database 'TestDB', file 'TestDB8' on file 1.
100 percent processed.
Processed 2 pages for database 'TestDB', file 'TestDB_log' on file 1.
BACKUP DATABASE successfully processed 19935482 pages in 387.441 seconds (401.986 MB/sec).
Backup Completed in 399 Seconds...
Technical white paper
Page 13
Database
: TestDB
Type
: DB_FULL
Name
: S=SKYSQL;I=DEFAULT;DB=TestDB;TS=2016-08-12T08:49:37.505;T=DB_FULL;V=1;ID=02C478B7-3379-4A0E-94886E7650BB3720;
Copied
: 155748.190 MB
Stored
: 47898.068 MB
Disk Space Saving : 69.246%
Bytes Transmitted : 47789.646 MB
Virtual Bandwidth : 390.346 MB/s
Actual Bandwidth : 119.774 MB/s
Bandwidth Saving : 69.316%
Mirror to
: Admin:admin@10.1.5.21/StoreOnce_SQL_catalyst2 (progress updated every 1 GB)
Completed in 428 Seconds...
Copied
: 15579.469 MB
Stored
: 48045.462 MB
Disk Space Saving : 69.152%
Bytes Transmitted : 47787.826 MB
Virtual Bandwidth : 363.901 MB/s
Actual Bandwidth : 111.654 MB/s
Bandwidth Saving : 69.318%
While the operations are running, the Activity Window on the first StoreOnce shows the device receiving the backup at more than 400 MB/s,
and then the Catalyst copy jobs replicating the backup to the second StoreOnce. The speed at which these operations are executed depends on
various factors, such as the network speed, the load on the StoreOnce, and the deduplication ratio (Chunks that are part of previous backups do
not need to be transmitted again.).
Technical white paper
Page 14
Figure 11. Monitoring performance using the StoreOnce appliance graphical interface.
Since the parameter maxstreams has not been specified in the command line, the plug-in automatically selected one stream or item per data file
in the DB, plus two control streams or items. These Catalyst items are all accessed in parallel during backups, restores, or Catalyst copies. These
Catalyst objects are visible on both the first and the second StoreOnce under the tab Catalyst Item Summary.
Technical white paper
Page 15
Figure 12. Items stored in an HPE StoreOnce Catalyst store, as viewed in the StoreOnce appliance graphical interface.
In addition to this, the backup to the first StoreOnce leaves traces in the tab Data Jobs; and the Catalyst copies leave trace in the tabs Outbound
Copy Jobs and Inbound Copy Jobs, respectively on the first and second StoreOnce.
Automating daily tasks
Daily operations generally include two tasks:
1. Regular backups—Typically, database backups are run once per day, and Transaction log backups are executed once every n hours.
2. Deleting expired backups—Usually, expired backups are deleted after a new successful backup.
These two tasks can be automated with scripts, scheduled at regular intervals using the Windows Scheduler. To write the scripts, it is
recommended to use PowerShell, which includes all the needed components:
• The assembly Microsoft.SqlServer.SMO can be used to enumerate the databases running under a specific server or instance.
• The SMO object pointing to a database has many properties that can be used to test if the database is accessible (.IsAccessible), is a system
database (.IsSystemDatabase), and requires a Transaction log backup (.RecoveryModel).
• PowerShell supports lists (collections) that can be used to manage include and exclude lists. For example, the database TempDB cannot be
backed up and should be part of the exclude list.
• Variables of type date have a property .ToString(‘dddd’) that can be used to identify the day of the week, allowing for example, a Full backup
on Friday and Differential backups on the other days of the week.
• The optional qualifier -tag=name should be used during backups so that backups of different types, for example, weekly and monthly, can
then be distinguished and expired with different retentions.
• To expire backups, the output of the -list –tag=name –date-end=date&time command should be processed, the backup names extracted,
and the backups deleted with the –delete command.
• The executable HPStoreOnceForMSSQL.exe returns a status code that can be tested with the variable $LASTEXITCODE.
Technical white paper
Page 16
Viewing backups on the Catalyst store
To see which backups exist on a specific Catalyst store, use the SQL Catalyst Plug-in with the –list option. For example:
C:\Program Files\Hewlett-Packard Enterprise\StoreOnce\isvsupport\sql\bin>HPStoreOnceForMSSQL.exe
-list -store="10.1.5.21/StoreOnce_SQL_catalyst1"
CT=2016-08-12 10:30:59; S=SKYSQL;I=DEFAULT;DB=TestDB;TS=2016-08-12T10:30:59.338;T=DB_DIFF;V=1;ID=61 B28AB2-87AC4ACB-B36C-E97D6E89E08B;
CT=2016-08-12 10:25:03; S=SKYSQL;I=DEFAULT;DB=TestDB;TS=2016-08-12T10:25:02.731;T=DB_FULL;V=1;ID=F2 3F6621-AAE943C1-A6B1-BE27BF25EBC8;
CT=2016-08-12 10:15:47; S=SKYSQL;I=DEFAULT;DB=TestDB;TS=2016-08-12T10:15:47.575;T=LOG;V=1;ID=837A0E 3C-8952-420FBB08-F36F7A50F629;
CT=2016-08-12 08:49:37; S=SKYSQL;I=DEFAULT;DB=TestDB;TS=2016-08 12T08:49:37.505;T=DB_FULL;V=1;ID=02
The previous example shows a log, Full and Differential backup of the TestDB database. The –database option can be used to show backups on
a store for a specific database. The list option can also be filtered to show backups by
• Instance name
• Database name
• Backup type
• Backup tag
• Date
• Status
In the following example and in Figure 13, only the backups for database LittleData are displayed.
C:\Program Files\Hewlett-Packard Enterprise\StoreOnce\isvsupport\sql\bin>HPStoreOnceForMSSQL.exe
-list -store="10.1.5.21/StoreOnce_SQL_catalyst1" -database LittleData
CT=2016-08-12 11:37:43; S=SKYSQL;I=DEFAULT;DB=LittleData;TS=2016-08-12T11:37:42.801;T=DB_FULL;V=1;I
D=DEF57D2E-8C65-46AD-9D21-8E2AA42FF4E3;
Technical white paper
Page 17
Figure 13. Backups stored in an HPE StoreOnce Catalyst store, as viewed in the StoreOnce Plug-in for SQL graphical interface.
Point-in-time restore
The database PRD needs to be restored to the state it had on 14 October at 3 a.m. Because we don’t want to overwrite the production database,
it needs to be restored under another name (P2), and new file locations have to be specified during the restore of the Full backup.
The Full, Differential, and Transaction logs backups can be restored using the following commands:
“C:\Program Files\Hewlett-Packard Enterprise\StoreOnce\isvsupport\sql\bin\HPStoreOnceForMSSQL.exe” -restore “RESTORE
DATABASE [P2] FROM STORE=‘192.168.1.133/TC-SQL/S=CBCSRV19;I=DEFAULT;DB=PRD;TS=2016-10-10T20:11:00.333;
T=DB_FULL;V=1;ID=F3A5086C-7F01-4B82-A9F1-B912525F0678;TAG=Weekly;’ WITH NORECOVERY, CHECKSUM, REPLACE, MOVE
‘PRD_01’ TO ‘M:\SQL\Data01\P2.mdf’, MOVE ‘PRD_02’ TO ‘N:\SQL\Data02\P2.ndf’, MOVE ‘PRD_log’ TO ‘L:\SQL\Log\P2.ldf’”
“C:\Program Files\Hewlett-Packard Enterprise\StoreOnce\isvsupport\sql\bin\HPStoreOnceForMSSQL.exe” -restore
“RESTORE DATABASE [P2] FROM STORE=‘192.168.1.133/TC-SQL/S=CBCSRV19;I=DEFAULT;DB=PRD;
TS=2016-10-13T20:11:52.358;T=DB_DIFF;V=1;ID=D50755E0-1D68-47EE-92AB-388BC645087F;TAG=Daily;’ WITH NORECOVERY,
CHECKSUM, REPLACE”
“C:\Program Files\Hewlett-Packard Enterprise\StoreOnce\isvsupport\sql\bin\HPStoreOnceForMSSQL.exe” -restore
“RESTORE LOG [P2] FROM STORE=‘192.168.1.133/TC-SQL/S=CBCSRV19;I=DEFAULT;DB=PRD;
TS=2016-10-13T22:06:39.692;T=LOG;V=1;ID=E50E39DB-6B49-4BE1-ACE9-A88DE5AC2018;TAG=Log;’ WITH NORECOVERY,
CHECKSUM, REPLACE, STOPAT=‘2014-10-14 03:00:00’”
“C:\Program Files\Hewlett-Packard Enterprise\StoreOnce\isvsupport\sql\bin\HPStoreOnceForMSSQL.exe” -restore
“RESTORE LOG [P2] FROM STORE=‘192.168.1.133/TC-SQL/S=CBCSRV19;I=DEFAULT;DB=PRD;
TS=2016-10-14T00:07:03.055;T=LOG;V=1;ID=5ACE6C32-A79D-4858-99B1-CCA2AEFD7F12;TAG=Log;’ WITH NORECOVERY,
CHECKSUM, REPLACE, STOPAT=‘2014-10-14 03:00:00’”
“C:\Program Files\Hewlett-Packard Enterprise\StoreOnce\isvsupport\sql\bin\HPStoreOnceForMSSQL.exe” -restore
“RESTORE LOG [P2] FROM STORE=‘192.168.1.133/TC-SQL/S=CBCSRV19;I=DEFAULT;DB=PRD;
TS=2016-10-14T02:06:29.646;T=LOG;V=1;ID=F68A430E-5112-4928-9295-7FE6807BA30A;TAG=Log;’ WITH NORECOVERY,
CHECKSUM, REPLACE, STOPAT=‘2014-10-14 03:00:00’”
Technical white paper
Page 18
“C:\Program Files\Hewlett-Packard Enterprise\StoreOnce\isvsupport\sql\bin\HPStoreOnceForMSSQL.exe” -restore
“RESTORE LOG [P2] FROM STORE=‘192.168.1.133/TC-SQL/S=CBCSRV19;I=DEFAULT;DB=PRD;
TS=2016-10-14T04:06:50.815;T=LOG;V=1;ID=45162641-EC80-48A5-A570-A947904BD2F3;TAG=Log;’ WITH RECOVERY, CHECKSUM,
REPLACE, STOPAT=‘2014-10-14 03:00:00’”
The steps required to generate these commands are
1. Identify all the backups needed for this restore with the command “-store=‘192.168.1.133/TC-SQL’ -list -database=PRD.”
The output lists, in descending order, the backups done on this DB. Select the Full, Differential (if any), and Transaction log backups required
to restore the DB to the selected time stamp. The backup name follows the Catalyst object creation time in each line of the output.
2. Use the plug-in and run a RESTORE FILELISTONLY on one of the backups of this database. The output contains the logical names of each
data and log file. These names are needed in the next step.
3. Restore the Full backup with a RESTORE DATABASE, specifying a MOVE TO clause for each file, and NORECOVERY to keep the database
open for further restores.
4. Restore the Differential backup with a RESTORE DATABASE and also with a NORECOVERY so that Transaction log backups can then be
applied.
5. Restore the Transaction log backups in the order they have been performed with a succession of RESTORE LOG. All but the last one needs
NORECOVERY; the last one requires RECOVERY.
Backup with Asynchronous Copy—Monitoring performance
The Asynchronous Copy option separates mirrored copies from the primary backup operations, which means that backups can complete faster,
and copies are scheduled in the background. Once a backup is completed, the copies are queued and control is returned to the plug-in GUI. The
copies from one StoreOnce appliance to another can now run independently of the backup jobs.
Note
This example includes extracts from the plug-in output that shows a very high deduplication ratio. This was achieved using backups where the
database had a very low change rate between backups. Should the database be subject of a large amount of changes, a lower deduplication ratio
and backup throughput may be observed.
The monitoring example uses the StoreOnce appliance GUI. To access this, the DBA requires user access rights to the StoreOnce appliance.
Technical white paper
Page 19
Figure 14. A policy responsible for Asynchronous Copy jobs, shown in the plug-in GUI.
“C:\Program Files\Hewlett-Packard Enterprise\StoreOnce\isvsupport\sql\bin\HPStoreOnceForMSSQL.exe -backup "BACKUP Database [TestDB]
TO STORE='Admin:admin@10.1.5.23/StoreOnce_SQL_catalyst5a' MIRROR TO STORE='Admin:admin@10.1.5.21/StoreOnce_SQL_catalyst5b'
WITH MAXTRANSFERSIZE=4194304, CHECKSUM,STATS=10" -async-copy”
Processed 2509184 pages for database 'TestDB', file 'TestDB' on file 1.
Processed 2508928 pages for database 'TestDB', file 'TestDB2' on file 1.
Processed 2508952 pages for database 'TestDB', file 'TestDB3' on file 1.
Processed 2509056 pages for database 'TestDB', file 'TestDB4' on file 1.
Processed 2509056 pages for database 'TestDB', file 'TestDB5' on file 1.
Processed 2509056 pages for database 'TestDB', file 'TestDB6' on file 1.
Processed 2509056 pages for database 'TestDB', file 'TestDB7' on file 1.
Processed 2509184 pages for database 'TestDB', file 'TestDB8' on file 1.
100 percent processed.
Processed 7 pages for database 'TestDB', file 'TestDB_log' on file 1.
BACKUP DATABASE successfully processed 20072479 pages in 376.180 seconds (416.864 MB/sec).
Backup Completed in 388 Seconds...
Technical white paper
Page 20
Database: TestDB
Type: DB_FULL
Name: S=SKYSQL;I=DEFAULT;DB=TestDB;TS=2016-08-19T10:04:32.670;T=DB_FULL;V=1;ID=7CC55FD
D-A792-42E3-AF4D-935B0D23C9C1;
Copied: 156820.190 MB
Stored: 48425.668 MB
Disk Space Saving: 69.120%
Bytes Transmitted: 48331.676 MB
Virtual Bandwidth: 404.176 MB/Sec
Actual Bandwidth: 124.566 MB/Sec
Bandwidth Saving: 69.180%
Mirror to: Admin:admin@10.1.5.21/StoreOnce_SQL_catalyst5b
Job Id: 441
Job Id: 442
Job Id: 443
Job Id: 444
Job Id: 445
Job Id: 446
Job Id: 447
Job Id: 448
Job Id: 449
Job Id: 450
Initiated Async Copy to Mirror:10.1.5.21/StoreOnce_SQL_catalyst5b
While the operations are running, the Activity Window on the first StoreOnce shows the device receiving the backup at more than 400 MB/s,
and then the Catalyst copy jobs replicating the backup to the second StoreOnce. The speed at which these operations are executed depends on
various factors, such as the network speed, the load on the StoreOnce, and the deduplication ratio (Chunks that are part of previous backups do
not need to be transmitted again.).
Technical white paper
Page 21
Figure 15. Monitoring performance using the StoreOnce appliance graphical interface for the backup and Asynchronous Copy job.
Since the Asynchronous Copy feature initiates the copy job once the backup is finished, the backup job performs similarly to a standard backup.
The Copy job also sees similar throughput to a Full backup because it’s also running separately from the backup job. The performance seen in
the GUI reflects this. The sporadic nature of the replication throughput is comparable to standard copy job.
Technical white paper
Page 22
Figure 16. Items stored in the HPE StoreOnce Catalyst stores, as viewed in the StoreOnce appliance graphical interface.
In addition to this, the backup to the first StoreOnce leaves traces in the tab Data Jobs; and the Catalyst copies leave trace in the tabs Outbound
Copy Jobs and Inbound Copy Jobs, respectively on the first and second StoreOnce.
Backup with Copy Redirect—Monitoring performance
The Copy Redirect feature allows the SQL Server Plug-in to have management access to the target StoreOnce appliance to set up copy operations,
while the copy itself (data transfer) occurs on a separate private network between the storage appliances. The copy address need not be accessible
to the Microsoft SQL Server and can be on a private network between the StoreOnce appliances such as the replication subnet.
Note
This example includes extracts from the plug-in output that shows a very high deduplication ratio. This was achieved using backups where the
database had a very low change rate between backups. Should the database be subject of a large amount of changes, a lower deduplication ratio
and backup throughput may be observed.
The monitoring example uses the StoreOnce appliance GUI. To access this, the DBA requires user access rights to the StoreOnce appliance.
Technical white paper
Page 23
Figure 17. Backups initiated with the Copy Redirect feature, shown in the plug-in GUI.
“C:\Program Files\Hewlett-Packard Enterprise\StoreOnce\isvsupport\sql\bin>HPStoreOnceForMSSQL.exe -backup” BACKUP Database
[TestDB] TO STORE='Admin:admin@COFC-wimbledon1-ssid2/StoreOnce_SQL_catalyst6b
' MIRROR TO STORE='Admin:admin@COFC-USE613C70101/StoreOnce_catalyst_phase6:16.78.1.8' WITH MAXTRANSF
ERSIZE=4194304, CHECKSUM,STATS=10"
Processed 2509184 pages for database ‘TestDB’, file ‘TestDB’ on file 1.
Processed 2508928 pages for database ‘TestDB’, file ‘TestDB2’ on file 1.
Processed 2508952 pages for database ‘TestDB’, file ‘TestDB3’ on file 1.
Processed 2509056 pages for database ‘TestDB’, file ‘TestDB4’ on file 1.
Processed 2509056 pages for database ‘TestDB’, file ‘TestDB5’ on file 1.
Processed 2509056 pages for database ‘TestDB’, file ‘TestDB6’ on file 1.
Processed 2509056 pages for database ‘TestDB’, file ‘TestDB7’ on file 1.
Processed 2509184 pages for database ‘TestDB’, file ‘TestDB8’ on file 1.
100 percent processed.
Processed 2 pages for database ‘TestDB’, file ‘TestDB_log’ on file 1.
BACKUP DATABASE successfully processed 20072474 pages in 399.026 seconds (392.997 MB/sec).
Backup Completed in 416 Seconds...
Technical white paper
Page 24
Database: TestDB
Type: DB_FULL
Name: S=SKYSQL;I=DEFAULT;DB=TestDB;TS=2016-08-22T09:37:10.031;T=DB_FULL;V=1;ID=743E680
8-C35C-4254-91A8-68F1E0821553;
Copied: 156820.190 MB
Stored: 48425.650 MB
Disk Space Saving: 69.120%
Bytes Transmitted: 48331.656 MB
Virtual Bandwidth: 376.972 MB/Sec
Actual Bandwidth: 116.182 MB/Sec
Bandwidth Saving: 69.180%
Mirror to: Admin:admin@COFC-USE613C70101/StoreOnce_catalyst_phase6:16.78.1.8 (progress updated every 1 GB)
Size: 156821 MB, Copied: 156821 MB, Speed: 351 MB/Sec, 0 Mins 0 Secs Left [**************]
Completed in 446 Seconds...
Copied: 156821.469 MB
Stored: 48583.390 MB
Disk Space Saving: 69.020%
Bytes Transmitted: 48329.003 MB
Virtual Bandwidth: 351.618 MB/Sec
Actual Bandwidth: 108.361 MB/Sec
Bandwidth Saving: 69.182%
While the operations are running, the Activity Window on the first StoreOnce shows the device receiving the backup at more than 400 MB/s,
and then the Catalyst copy jobs replicating the backup to the second StoreOnce. The speed at which these operations are executed depends on
various factors, such as the network speed, the load on the StoreOnce, and the deduplication ratio (Chunks that are part of previous backups do
not need to be transmitted again.).
Technical white paper
Figure 18. Monitoring performance using the StoreOnce appliance graphical interface for backup during the Copy Redirect job.
Figure 19. Monitoring performance using the StoreOnce appliance graphical interface for copy during the Copy Redirect job.
Page 25
Technical white paper
Page 26
Figure 20. Items stored in the HPE StoreOnce Catalyst stores, as viewed in the StoreOnce appliance graphical interface.
Figure 21. Items stored in the HPE StoreOnce Catalyst stores, as viewed in the StoreOnce appliance graphical interface, on the Copy Target machine.
Performance
Performance (backup speed, restore speed, and deduplication ratio) depends on many factors such as available resources on the SQL Server,
network load, number of concurrent backups, amount changed in the DB between two backups, retention period, and so on.
HPE testing has shown that the HPE StoreOnce Plug-in for Microsoft SQL Server has been seen providing better performance than alternative
solutions in the vast majority of use cases. This point is further illustrated by a test conducted under the following conditions:
• 500 GB database with eight data files, each on a different mount point
• 60 Full database backups with 30 backup retention
• Approximately 5% of the DB’s rows or bytes changed between the backups
Comparing the performance of the plug-in vs. backups to a StoreOnce CIFS share, we found the HPE StoreOnce Plug-in for Microsoft SQL Server
providing 18.6% faster backups and saving 2.8 times more disk space.
Technical white paper
Page 27
Deduplication ratio
Backup speed (MB/s)
50
45
40
35
30
25
20
15
10
5
0
520
500
480
460
440
420
400
380
Plug-in for SQL
CIFS
Figure 22. Measured average backup throughput over 60 Full backups.
Plug-in for SQL
CIFS
Figure 23. Store deduplication ratios after two months of backup and one-month retention.
Note
Approximately 40% of the maximum throughput of a 10 Gbps Ethernet connection is required to achieve 450 MB/s to a CIFS share. In other
words, the Ethernet connection becomes rapidly the bottleneck when several MS SQL Servers or databases are backed up in parallel.
The plug-in for MS SQL only consumes a small percentage of the throughput; therefore, it becomes a lot faster when multiple SQL Servers or
databases are backed up in parallel.
Low bandwidth vs. high bandwidth
Hewlett Packard Enterprise recommends using low-bandwidth mode on Catalyst stores used with the HPE StoreOnce Plug-in for MS SQL Server.
Using low-bandwidth mode delivers significantly faster backup and improved space savings for many use cases, as shown in Figures 24 and 25.
Figure 24. Low-bandwidth vs. high-bandwidth throughput with 1% data change.
Technical white paper
Page 28
Figure 25. Low-bandwidth vs. high-bandwidth deduplication ratio with 1% data change.
Data change rate
Data change rate can have a significant impact on deduplication ratios. Deduplication ratios will be higher if the data change rate is low. The
following chart illustrates how the database change rate between each Full backup affects StoreOnce deduplication ratios.
Figure 26. Effect of various data change rates on deduplication ratio through two weeks of Full backups.
Technical white paper
Page 29
Backup retention period
Backup retention periods can also have a significant impact on deduplication ratios. Deduplication ratios will be higher if the retention period is long.
In the following chart, the retention period was 30 days with a Full backup every day. Once the retention period is reached, a decline in deduplication
ratio of about 10% is expected.
Figure 27. Effect of retention period on deduplication ratio—30-day retention period and 1% data change rate was used.
Note
There will be a slight decline in deduplication ratio once the retention period is reached; however, the deduplication ratio will stabilize. The reason
for this decline is that the backup being expired has a small amount of more deduplicated data than the newer backup. This phenomenon is not
a characteristic of the HPE StoreOnce Plug-in for SQL Server and is not data type dependent.
Known limitations
Please note the following known limitations of the HPE StoreOnce Plug-in for Microsoft SQL Server v1.2:
• No support for FILE and FILEGROUP backups.
Technical white paper
Page 30
Summary
The StoreOnce Plug-in for Microsoft SQL Server in conjunction with Catalyst targets on StoreOnce appliances is the highest performing solution to MS
SQL Server backups as the test results in this paper demonstrate. Compared to existing backup solutions, it gives the Database Administrator increased
control over how backups are run and the ability to finish backups faster while using less space to store the backups. All backup and restore operations
are executed using commands similar to known T-SQL commands. In summary, the reasons to deploy the StoreOnce SQL Server Plug-in are:
• Improved backup throughput compared to back up to CIFS targets:
– Due to the enhanced efficiency of the Catalyst protocol
– By distributing deduplication processing to the MS SQL Server(s) rather than having the StoreOnce Backup target executing all the
deduplication
• Reduced space consumed by database backups because of higher deduplication ratios achieved by the intelligent processing of the backup
stream
• Lower network bandwidth use achieved with low-bandwidth backups that only transmit the unique data to the StoreOnce appliance
• Added protection and DR capability with up to three additional copies of the backup made using MS SQL MIRROR TO directive
• No learning curve as most of the standard Microsoft SQL Server BACKUP/RESTORE WITH options supported
• Zero cost for the plug-in that is downloadable from HPE Software Depot
Appendix A—HPE StoreOnce Catalyst overview
Supported device types
HPE StoreOnce Backup appliances can be configured with three types of devices that can be used as backup targets. Multiple device types can
be used in parallel to support different workloads. The choice of device type is determined by the needs of the data protection solution being
deployed.
• Virtual tape library (VTL) is supported for connection over Fibre Channel and Ethernet (using iSCSI) networks. This is used by traditional backup
software that was first designed for use with tape-backed devices. This device type emulates the geometry of physical tape devices for simple
integration with backup software. It is robust and well proven but provides limited advanced functionality compared to StoreOnce Catalyst devices.
• NAS (CIFS and NFS) devices are supported for connection over Ethernet networks. These device types are primarily used for newer backup
applications that were not initially designed for use with tape devices. NAS is also used for business applications, such as Microsoft SQL Server,
which have built-in backup and recovery modules.
• HPE StoreOnce Catalyst is a device type specifically designed for backup and recovery. Catalyst is supported for connection over Fibre Channel
and Ethernet networks. It removes many of the limitations of the general purpose VTL and NAS device types and adds features for enhanced
control and performance. It uses a client-server architecture; the Catalyst server runs on the StoreOnce appliance, and the Catalyst client is
integrated with the backup or business application.
StoreOnce Catalyst terminology and summary features
HPE StoreOnce Catalyst store is the target device where backups are stored.
HPE StoreOnce Catalyst items are objects not files. Backup data is held as Catalyst items in a Catalyst store. Item size and number are set by
the StoreOnce Plug-in for Microsoft SQL Server.
An HPE StoreOnce Catalyst client, such as the StoreOnce Plug-in for Microsoft SQL Server, communicates with the StoreOnce Catalyst server
that runs on the StoreOnce appliance.
A Catalyst data job is any job that stores or restores data to or from a Catalyst store.
A Catalyst copy job copies backup data between Catalyst stores. The StoreOnce Plug-in for Microsoft SQL Server is aware of these copies and
all can be used for restore jobs.
The design of the Catalyst device type enables many advanced features.
Technical white paper
Page 31
Source-side deduplication can be configured for low-bandwidth backups to help reduce, usually by 90% or more, the load on the network by only
transmitting new and updated blocks to complete the backup. If there are multiple application servers or media servers running the Catalyst client
software, it means the deduplication is federated and distributed so it is possible to achieve overall higher throughput to the StoreOnce appliance for
reduced times to complete backups.
HPE StoreOnce Catalyst also enables more flexible replication—known as Catalyst copy. NAS and VTL replication is a 1:1 asynchronous mirror
between the source and target. Catalyst copies enable added features such as multihop replication.
Integrating the Catalyst client with the backup software or business application enables control of the StoreOnce capabilities to be executed
through the backup software or business application for a unified point of backup management. Depending upon the level of integration, this
enables the application to manage device configuration, backup specification, copy configuration, and view reporting and troubleshooting
information. Figure 28 illustrates the use of StoreOnce appliances in a data protection solution that includes backups and replicated backups.
Figure 28. Multiple HPE StoreOnce device types in use.
The HPE StoreOnce Plug-in for Microsoft SQL Server fully supports CoFC. When this protocol is used, the data is transferred using the Fibre
Channel protocol instead of the Ethernet.
Technical white paper
The HPE StoreOnce Plug-in for Microsoft SQL Server also supports Federated Catalyst stores, with the restrictions that such a store cannot be
the target of a Catalyst Inbound Copy Job. In other words, it can be used in a TO STORE expression but cannot be used in a MIRROR TO STORE
clause.
Useful links
HPE StoreOnce official page
HPE StoreOnce appliance manuals
HPE StoreOnce Catalyst Solution Service data sheet
Microsoft TechNet for SQL Server
To get your copy of the StoreOnce Plug-in for Microsoft SQL Server
1. Go to HPE Software Depot.
• Expand the Storage menu and select HPE StoreOnce.
2. Under HPE StoreOnce, locate and select StoreOnce Application Plug-ins for Direct Backup.
3. Locate and read the information about the HPE StoreOnce Plug-in for Microsoft SQL Server.
4. When prompted, enter your HPE Passport credentials (if not sign up—it is free and fast).
5. Check your details on the next page and read license terms, then select Next.
6. From the Software downloads and licenses page, select the StoreOnce Plug-in for Microsoft SQL Server and select Download.
7. Follow instructions to download the package containing the plug-in and the user guide.
Note
By default, the StoreOnce Plug-in for Microsoft SQL Server user guide is installed in \Program Files\Hewlett-Packard
Enterprise\StoreOnce\isvsupport\sql\doc.
Learn more at
hpe.com/storage
Sign up for updates
© Copyright 2014–2016 Hewlett Packard Enterprise Development LP. The information contained herein is subject to change without
notice. The only warranties for Hewlett Packard Enterprise products and services are set forth in the express warranty statements
accompanying such products and services. Nothing herein should be construed as constituting an additional warranty. Hewlett Packard
Enterprise shall not be liable for technical or editorial errors or omissions contained herein.
Microsoft, Windows, and Windows Server are either registered trademarks or trademarks of Microsoft Corporation in the United States
and/or other countries. All other third-party trademark(s) is/are property of their respective owner(s).
4AA5-6133ENW, October 2016, Rev. 2
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF

advertising