Several data paths can be added to a storage policy copy, to ensure the
success of data protection and other operations conducted using the storage
policy. A data path is the combination of MediaAgent, Library, Drive Pool and
Scratch Pool used by the storage policy copy to perform a data protection
operation. Each storage policy copy has a default data path which will be used
to perform data protection operations. In addition, you
can also define alternate data paths in each of the storage policy copies.
Alternate data paths provide the following advantages:
Alternate data
paths provide the facility to automatically switch over to an alternate data path, when one
of the components in the default data path is not available. In addition to
ensuring the successful completion of data protection jobs, alternate data
paths also utilize available libraries and drives in the event of failure or
non-availability of these resources.
Alternate data paths can be used to minimize media utilization by
routing data protection operations from several subclients to the same
storage policy and hence the same media, instead of creating several storage
policies which in turn utilizes a different media for each subclient.
In addition, the facility to load balance (round robin) between alternate data paths
provides the mechanism to load balance or evenly distribute data
protection operations between available resources.
Alternate data paths are supported for both the primary and secondary copies
associated with storage
policies for all libraries. (See Alternate
Data Paths (GridStor) -Support
for additional details.) Note, however, that there are several differences
between the operations performed using primary and secondary copies with
alternate data paths. These are explained in detail in the following sections.
Note that within the selected storage policy (and its data paths), the
facility to define a subset of the data paths at the subclient level is also
provided.
If
a storage policy is created during the library configuration process, a default data path is created for the
primary copy using the MediaAgent, Library, Drive Pool and default scratch pool combination for drive
pools configured within the library. If you create a new storage policy, you
must specify a Library, MediaAgent, Drive and Scratch pool combination for the primary
copy.
Additional data paths for the primary copy can be defined from the
Data
Paths tab of the Copy Properties dialog
box.
The data paths that are available to be added as alternate data paths,
depends on the option selected in this dialog box. In addition some of the
options may require the index cache to be shared to be accessed as a data path.
(These details are explained in the subsequent sections of this document.)
After defining additional data paths, if necessary, you can set any of the data paths as
the default data path for the storage policy copy.
When multiple MediaAgents share the same library (SAN DDS, or direct-attached
shared library configurations) the system can automatically add the alternate
data paths for each of the storage policies, when this option is enabled. As
each of these data paths (MediaAgent, Library, Drive Pool and Scratch Pool) use
the same resources, additional index cache configuration is not required. In
addition, the criteria for using the alternate data path (described in the
following section) must also be specified.
A storage policy can be configured to use an alternate data path using the
following criteria:
When resources are busy or offline - use this option to configure your
system to use an alternate data path when resources are busy or offline.
Load balance between the data paths - use this option to evenly distribute
data protection operations amongst drive-pools, thereby not overloading a
specific drive-pool.
LAN preferred data path - use this option to automatically perform
LAN-free data protection operations.
When this option is selected the system automatically uses an alternate data
path when resources are busy or offline.
If the When resources are offline option is selected, the storage policy will use an alternate data
path when one of the following resources is broken or not available and
hence marked as offline by the user or by the
MediaAgent:
MediaAgent
Library
Master Drive Pool
Drive Pool
All the drives in the Drive Pool
No spare media in the scratch pool associated with the copy
If the When resources are busy option is
selected, the storage policy will use an alternate data path when all the drives
in the library are busy.
In both the above options, you can indicate whether an alternate data path must
be used immediately or after the specified amount of time.
The list of data paths that will be available when this option is selected
will include the following:
Data paths associated with MediaAgents that share the library with the
default data path. In this case, it is not necessary to share the index
cache, but the number of alternate data paths will be limited to the list of
MediaAgents that share a library.
List of MediaAgents that share the index with the MediaAgent associated
with the default data path. In this case, as other libraries can be included
in the list of data paths several alternate data paths can be added.
However, keep in mind that the index cache must be shared.
See Index Cache for information on the options available for index cache sharing.
When this option is selected the system automatically performs load-balancing
between the resources (drives in a library or writers in a disk library)
available in all the data paths. Keep in mind that the load-balancing is
performed at the drive-pool level as opposed to the MediaAgent level.
The
following section illustrates the load balancing operation:
If you have defined
5 data paths with 15 resources, and have 25 data protection operations running
concurrently at a given time, load balancing would cause the following to occur:
The first operation will be performed on the default data path.
The second (and subsequent operations) will be performed in the next data
paths, in the order in which it is added in the Data
Paths tab of the Copy Properties dialog
box.
Once the first 5 operations reserves the resources, the sixth operation
will be routed to another resource in the default data path, if one is
available. The subsequent operations will be routed to the next data path in
the order in which it is added in the Data Paths
tab of the Copy Properties dialog box, until all
the resources are occupied.
Once all the resources are used, the system will constantly check for
an available resource, and as soon as one is freed the next job in the queue
will be automatically routed to use that resource.
All the MediaAgents that share the index cache with the MediaAgent in the
default data path will be available as an alternate data path when this option
is selected.
When this option is selected the system automatically performs LAN free
backups wherever possible. It is not necessary to share the index cache for this
operation and all available MediaAgents will be available as an alternate data
path when this option is selected.
When you add or delete an alternate data path, you must reset the number of
streams that are defined for the Storage Policy.
The maximum number of streams for a storage policy, with a primary copy that
has alternate data paths should be equal to the sum of all unique drives
associated with the drive pools and/or the sum of all writers in the mount paths
associated with disk libraries in all alternate data paths. Consider
the following scenarios, when the maximum number of streams defined is either
too many or too little, when you have specified the criteria to immediately use
alternate data paths when resources are busy:
If the maximum number of streams in a Storage Policy is less than the sum
of drives/writers in mount paths associated with all the data paths in
a primary copy, then all the resources (drives/writers in mount paths)
available in the data paths will not be utilized. For example, if the sum of
drives/writers in mount paths in all the data paths is 20, and you have
specified 10 as the maximum number of streams, at any given time, only 10
jobs would succeed and the remaining jobs would go into the Waiting
status with the Job Delay Reason stating that no
resources are available for the job. In such a situation to fully utilize
all the available resources, the maximum number of streams should be set to
20.
If the maximum number of streams in a Storage Policy is more than the sum
of drives/writers in mount paths associated with all the data paths in a
primary copy, only as many jobs as the total number of available drives will
succeed. For example, if the sum of drives/writers in mount paths in all the
data paths is 20, and you have specified 30 as the maximum number of
streams, at any given time, only 20 jobs would succeed and the remaining
jobs would go into the Waiting status with the Job
Delay Reason stating that no resources are available for the job.
Jobs with Multiple Streams
For multi-stream jobs, the failover will occur only when all the streams have
the necessary resources. For example, if you have a job with 5-streams, and if
the necessary resources are not available in the default data path, the failover
will occur only when the alternate data path has all the necessary resources -
MediaAgent, Library and drive pool with 5 drives. This is the case, irrespective
of the criteria (When resources are busy, Round Robin between Data Paths options) specified to use
alternate data paths.
Data paths can be added to secondary copies to enable LAN free Auxiliary Copy
operations, so that network resources can be freed wherever possible.
Adding Data Paths to Secondary Copies
When a
secondary copy is created, you must select the default data path for the copy by
selecting the MediaAgent, Library, Drive Pool and scratch pool combination. This
data path will be used to access the secondary copy when an Auxiliary Copy
operation is performed.
However you can add data paths for the secondary copy so that any Auxiliary
Copy operations can be performed using a LAN-free data path.
As with the primary copy, additional data paths for the secondary copy can
also be defined from the Data
Paths tab of the Copy Properties dialog
box. (See
Add a Data Path to a Storage Policy Copy for step-by-step instructions.)
Note that although the Use preferred datapaths
option is selected, the LAN free Auxiliary Copy operations on the copy is
not performed until the alternate data paths are selected.
When you add the data path for the secondary copy it is sufficient to add one
path per MediaAgent-Library combination. The system automatically uses an
available data path to perform LAN free Auxiliary Copy operations. Keep in mind,
that when you add data paths in the secondary copies, the system automatically
tries to perform a LAN-free read operation. (This is opposed to the
primary copies, where the system strives to perform both the read and write
operations when the LAN-free option is selected.)
Each subclient can be configured with a subset of data paths from the data paths
available in the storage policy associated with the subclient . The following options are provided while
defining the data paths for a subclient:
Facility to select a subset of data paths from the list of available
data paths.
Facility to assign a priority for the selected data paths.
If necessary, facility to override the default data path on storage
policy and use the other data path from the subset of data paths available
for the subclient.
Note that Override Datapaths option is
not supported if the subclient is associated with an Incremental Storage Policy.
Note that the data paths and the priority established at the subclient level
takes precedence over the data paths defined at the storage policy copy.
By default, the system uses the data path associations defined in the
primary copy of the storage policy to perform data protection
operations. (This is depicted in the sample image shown on the right.
Note that the Override Datapaths option
is not selected and the default data path is displayed using a bold
font-face and a special icon.) If necessary you can perform the
following operations:
Select a subset of the available data paths for the subclient.
Set a priority for the selected data paths.
See the following procedures for step-by-step instructions:
Consider the following for configuring data paths at the subclient level:
Subclient data paths are supported by all agents that require an
iDataAgent Backup Storage Policy.
If necessary, different data path subsets can be selected for database
agents that use different storage policies for data and log files. (See,
Classification of Agents based on Index Usage for a definition of
database agents.)
When a secondary (storage policy) copy is promoted as the primary copy,
the data paths defined in the secondary copy is automatically used by the
subclient. If necessary, you must establish the default data path and
set the priority once the secondary copy is promoted.
When a data path is deleted from the storage policy (or a library is
deconfigured) the data path is automatically removed from the subclient.
Subclient data paths are not supported by Subclient Policies.
Subclient data paths cannot be configured using the Command Line
interface.
In the case of
Incremental
Storage Policies, which uses two different storage policy copies for
full and non-full backups, different data path subsets can be selected for
full or non-full backups using the same or a different storage policy. In
such a situation, make sure that the selected data paths share the index.
The subclient data paths can be used effectively in the following situations:
To configure subclients to use certain data paths, and minimize media
utilization. See Example 1 for
information on how this works.
Restrict some subclients to use only a subset of data paths and at the
same time reduce media utilization. See
Example 2 for information on how this works.
When a data protection operation is initiated, the storage policy copy
attempts to write the data using the default data path. If the default data path
is not available, an alternate data path is automatically used to perform the
data protection operation. If more than one alternate data path is defined, the
first data path listed in the Data
Paths tab of the Copy Properties
dialog box is selected, followed by the second and so on until a data path is
available.
If both the default and alternate data paths are configured to use the same
library, as a result of a shared library configuration (configured as a SAN DDS
library or direct-attached shared library) the MediaAgent will automatically
use the appropriate Assigned media for the data
protection operation.
If the default and alternate data paths are configured to use different
libraries, the MediaAgent, marks the previously used Assigned
media as Appendable and uses a new media from the
library associated with the alternate data path.
Such Appendable media can be re-used in the
library by enabling the Use Appendable Media option
in the Library
Properties dialog box associated with the library.
When a Data Recovery operation is initiated, and if the media is not
exported, the software attempts to restore/recover data using the
appropriate data path associated with that library, instead of the default data
path in the following order:
The first priority is provided to the path which results in LAN-free
restore/recover to the client computer from which the restore/recover
operation was initiated. LAN-free operation is possible only when the client
initiating the restore/recover operation and MediaAgent are on the same
computer.
If the LAN-free operation is not possible, then the Data Recovery
operation attempts to restore/recover data using the default data path.
If the appropriate media is not available in all these data paths, the
software automatically identifies the media in which the data resides
and performs the restore/recover operation from that library.
When a Data Recovery operation is initiated, and if the media is exported,
the software will prompt you to import the media in the appropriate
MediaAgent computer. This is done as follows:
If a LAN-free restore/recover is possible, the restore/recover operation
would prompt you to import the media in the appropriate library from which the
LAN-free restore is possible. (LAN-free restore will be possible only when the
client initiating the restore/recover operation and MediaAgent are on the same
computer.)
If the LAN-free restore/recover operation is not possible, then the operation
would prompt you to import the media in the library which was last used to write
to the media.
If the resources in that library are offline, the restore/recover operation
would prompt you to import the media in the library associated with the default
data path.
If the resources associated with the default data path are offline, then the
restore/recover operation identifies a library from an alternate data path which
are assigned in the data path list, to import the media.
Job Preemption is supported on Default/LAN-free backups (i.e., if GridStor
is configured for User Preferred Datapath). It is not supported for the jobs
that are run using data paths with the Round Robin configuration.
Change Data Path (right-click option) should not be performed on Storage Policies with
Storage Policy Copies that have Alternate Data Paths (GridStor). If Change Data
Path is performed on such a setup, data recovery operations can be
performed from the media. However, subsequent data protection operations will
not re-use the migrated media.
On clustered computers the system automatically performs LAN free
operations for Agents installed on the virtual machines with the storage
policy copy (attached to the subclient) pointing to the MediaAgent on the
physical node. Consider the following example:
A file system
iDataAgent is
installed on Virtual Machine (VM1) and can be controlled by Node 1 or Node 2
at any given time.
The subclient (subclient1) associated with this file system
iDataAgent on VM1 points to a
Storage Policy Copy (SP1) which in turn uses the following data paths:
default data path using MediaAgent (Node1) and Library 1
alternate data path MediaAgent (Node2) and Library 1
When a backup is run on subclient1, the system automatically figures out
the node controlling VM1 and will use the appropriate MediaAgent. For
example if VM1 is controlled by Node 2 at the time of the backup, the system
automatically uses the MediaAgent on Node2 to perform the LAN free backup.
This capability allows you to install the MediaAgent on the physical node
of a cluster. (Instead of multiple instances if installed in the cluster
group.) However you will need GridStor® to provide failover
capabilities.
Data paths for NAS attached libraries can only be added if the MediaAgent
used in that data path also has the File System
iDataAgent installed on that computer. This is applicable only for
Windows MediaAgents.
If the MediaAgent software is installed in the cluster server, configure
a disk library to backup the data.
If you wish to configure a tape/optical library, install the MediaAgent
software on the physical computer.
Add a data path which uses this
MediaAgent, Library, Drive Pool and Scratch Pool combination to the Storage
Policy used to backup the cluster server. (See
Add a Data Path to a Storage Policy Copy for step-by-step instructions.)
Assign this as a high priority data path in the subclient(s) used to
backup the cluster server. (See
Assign Priorities for
Subclient Data Paths for step-by-step instructions.)
When the devices are configured from different HBA cards on the same host as
distinct drive pools, each of these drive pools can be added as data paths on
the storage policy. For LAN storage policies, these data paths can be added as
additional data paths with the Round-Robin between Data Paths option
enabled. This will automatically round robin the data protection jobs between
these HBA cards.
For LAN free Storage Policies, the additional LAN free data paths for the
secondary HBA cards can be added as data paths and the software will
automatically pick the least used LAN free data path.
When a storage policy is configured for multiplexing and contains data
paths to be used in round-robin fashion, then the round-robin option is completely
utilized before using multiplexing.
For example, consider a storage policy with 3 data paths configured in the
round robin mode and multiplexing set to 2.
If 4 backup jobs kick off simultaneously, then the three available data paths
are utilized first and multiplexing is applied for the fourth job only.
Consider the following information and recommendations, while creating and
using alternate data paths:
For LAN-free clients do not enable the When Resources are Busy option
to choose an alternate data path. This will ensure LAN free operations,
wherever possible.
It is not necessary to share the index on MediaAgents with LAN-free data paths. However, even if one additional alternate data path on
the LAN is added to the storage policy, you must share the index for all the
MediaAgents in the data path list.
Create and use less number of storage policies, as large number of storage
policies will result in the fragmentation of data on media. Consider the
following:
Control data retention by creating copies within the Storage Policies.
(As opposed to creating many storage policies with different retention
periods.)
Before you deconfigure a library, verify and ensure that none of the Storage
Policy Copy's default data path points to the library. (See
View the Storage Policies
Accessing a Library
for step-by-step instructions on how to view the storage policies associated
with a library.) If necessary, set an alternate data path as the default data
path before deconfiguring the library.
If you have a storage policy copy with no default data path,
use Change Data Path option to migrate the storage policy to point to
another data path. See Change Data Path for
more information.
NAS Load-Balancing - in addition to the resource load
balancing that the Alternate Data Paths feature provides, NAS can be configured to load balance the processing tasks associated with
backup, restore, and auxiliary copy jobs, which normally run on a single client
machine, to spread the processing among different MediaAgent.
If
a storage policy is created during the library configuration process, a default data path is created for the
primary copy using the MediaAgent, Library, Drive Pool and default scratch pool combination for drive
pools configured within the library. If you create a new storage policy, you
must specify a Library, MediaAgent, Drive and Scratch pool combination for the primary
copy.
