Previous  |  Next  >  
Product: Storage Foundation Guides   
Manual: Storage Foundation 4.1 Intelligent Storage Provisioning Administrator's Guide   

ISP Concepts

The figure, Relationship Between Concepts in ISP, illustrates the dependencies between the various ISP components.

Relationship Between Concepts in ISP

Relationship Between Concepts in ISP

Click the thumbnail above to view full-sized image.

These concepts are defined in the following sections:

Disk Group

A disk group is a named collection of disks that share a common configuration. Volumes and other VxVM objects must be created within a disk group, and are restricted to using disks from within that disk group.

LUN

A LUN, or logical unit, can either correspond to a single physical disk, or to a collection of disks that are exported as a single logical entity, or virtual disk, by a device driver or by an intelligent disk array's hardware. VxVM and other VERITAS software modules may be capable of automatically discovering the special characteristics of LUNs, or you can use the annotation service, provided in the VEA graphical user interface, to define new storage attributes.

Attribute

A storage attribute allows the properties of a LUN to be defined in an arbitrary conceptual space. For example, attributes can describe properties such as:

  • Disk access name
  • Disk media name
  • Manufacturer
  • Model type
  • Physical location, such as rack number, frame number, floor, building, or site
  • Hardware RAID configuration
  • Failover properties
  • Performance properties

  • Note   Note    You can use the annotation service, provided in the VEA graphical user interface, to create and delete storage attributes.

Attributes may be used to capture information about special features that storage possesses, such as:

  • Hardware-supported cloning, such as EMC Business Continuity Volumes (BCV)
  • Hardware-supported replication, such as the EMC Symmetrix Remote Data Facility (SRDF)
  • Hardware redundancy, such as mirrored parity, and caching

It should only be necessary to enter such information manually if VxVM cannot discover it automatically. An example of a user-defined attribute is physical location.


Note   Note    Attribute names and their string values are case sensitive. You can use the vxvoladm listattrs command to discover the correct spelling of LUN attribute names.

Storage Pool

A storage pool is defined within a disk group in VxVM for use by ISP. A storage pool is a policy-based container for LUNs and volumes. This means that the templates, capabilities and policies that are associated with a storage pool define how storage is organized within the pool.

For convenience, storage pool definitions are provided that include a number of associated templates that can be used for different purposes. See Storage Pools for details of the storage pool definitions that are provided.

Two types of storage pool are defined: data pools and clone pools. These are described in the following sections.

Data Pool

The first storage pool that is created within a disk group. All other storage pools that are subsequently created within a disk group are clone pools.


Note   Note    Only one data pool can be created within a disk group.

It should not usually be necessary to move a data pool to another disk group. However, if this is required, only an entire data pool can be moved. An individual application volume within a data pool cannot be moved. If you want to move a data pool into a different disk group, you must ensure that the data pool contains only those objects that you require.

Clone Pool

A storage pool that contains one or more full-sized instant volume snapshot replicas of volumes within a data pool. (A volume snapshot is an image of a volume's contents at the moment in time that the snapshot was taken. See the VERITAS Volume Manager Administrator's Guide for more information about volume snapshots.)


Note   Note    Only full-sized and space-optimized instant snapshots are supported for use with ISP. Third-mirror break-off volume snapshots are not supported.

A clone pool can be moved into another disk group, but an individual snapshot within it cannot. If you want to move a clone pool into a different disk group, you must ensure that the clone pool contains only those snapshots that you require.

Storage Pool Set

A storage pool set is a bundled definition of the capabilities of a data pool and its clone pools. For convenience, you can use a storage pool set definition to define both pools in a single operation. See Storage Pool Sets for details of the storage pool sets that are provided.

Policy

A storage pool's policies define how it behaves when more storage is required, and when you try to create volumes whose capabilities are not permitted by the current templates. The two policies that are associated with a storage pool are:

AutoGrow Defines how a storage pool uses LUNs. This policy can take the following level values:

    pool Use storage that has been manually assigned to the storage pool.

    diskgroup Use storage that has been assigned to the disk group.

SelfSufficient Defines how a storage pool uses templates. This policy can take the following level values:

pool Use templates that have been manually assigned to the storage pool.

    diskgroup Use templates that have been manually assigned to the disk group.

    host Use templates that have been installed in the ISP repository on this host.

The values of these two policies can be combined to suit how the storage pool is to be used.

The policy levels that are associated with a storage pool control how it manages its templates and LUNs:

  • The default level of the AutoGrow policy is diskgroup. This allows LUNs that are associated with the storage pool and its disk group to be used for allocating storage to volumes.
  • The default level of the SelfSufficient policy is pool. This allows only templates that are associated with the storage pool to be used for allocating storage to volumes. The storage pool can contain only volumes with the reliability and performance capabilities that are supported by these templates.

The scope of the levels that can be configured as policy values for a storage pool is illustrated in Scope Levels for Storage Pool Policies.

Scope Levels for Storage Pool Policies

Scope Levels for Storage Pool Policies

Click the thumbnail above to view full-sized image.


Note   Note    LUNs in other storage pools or disk groups or outside any disk group are not immediately available for aggregating into another storage pool or disk group.

Capability

A capability is a feature that is provided by a volume. For example, a volume may exhibit capabilities such as performance and reliability to various degrees. Each type of capability is defined as a set of rules.

The following table shows some simple examples of capabilities that might be supported by a storage pool:

Capability Adjustable Parameters Description

DataMirroring

nmirs --- number of mirrors (plexes)

Provides reliability by using mirrored plexes.

Raid5Capability

ncols --- minimum number of columns

nmaxcols --- maximum number of columns

Provides reliability by using RAID-5 parity.

Raid5LogMirroring

nlogs --- number of log copies

Provides reliability for RAID-5 logs by mirroring.

Striping

ncols --- minimum number of columns

nmaxcols --- maximum number of columns

Provides performance by striping across columns.

Capabilities have variable parameters that you can specify, such as the number of mirrors or columns. You can adjust the values of these parameters to tune the characteristics of a volume. If you do not specify parameter values, default values are used.

See Capabilities for details of the capabilities that are provided.

How capabilities are used to specify performance and reliability characteristics is described in the following sections.

Performance

Performance specifies the capabilities of a volume based on factors such as number of columns in striped volumes, stripe unit size, preferred characteristics of storage such as using prefabricated disks that are configured in hardware.

Reliability

Reliability specifies the level of redundancy that is required from a volume as a capability. Very reliable volumes have a high degree of redundancy. For example, a very reliable volume could be configured as a software mirrored volume built upon underlying prefabricated RAID-5 capable LUNs that are configured in the enclosure hardware. A medium-level redundant volume could be a simple 2-way mirror or RAID-5 volume configured either in software or on a suitable LUN within a single enclosure.

Rule

A rule is a statement written in the VERITAS ISP language that specifies how a volume is to be created. A rule can define selection of storage or layout of storage. Rules are usually gathered together as templates for creating volumes, rather than being specified individually.

Examples of storage selection and storage layout rules are described in the following sections.

Storage Selection Rules

Storage selection rules specify what storage can be used to create volumes. The following are examples of storage selection rules:

Rule Description

affinity

Attempts to select storage with shared attributes.

confineto

Restricts selection of storage by specifying attributes.

exclude

Prevents storage with certain attributes from being selected.

select

Specifies the storage to be used.

separateby

Defines fault domains for objects such as plexes.

strong separateby

Does not allow objects in different fault domains to share attributes.

For example, the following rule specifies that any LUNs may be selected from the named enclosures:


select "Enclosure"="ENC01", "Enclosure"="ENC02"

The next rule specifies that LUNs may be selected from the set of Hitachi disks that are located in Room1:


confineto "VendorName"="HITACHI", "Room"="Room1"

This example of a separation rule specifies that enclosures are to be treated as individual fault domains:


separateby "Enclosure"

Storage Layout Rules

Storage layout rules specify how storage is used to create volumes. The following are examples of storage layout rules:

Rule Description

log

Specifies the type of log and its degree of redundancy.

mirror

Specifies how many mirrors a volume should have.

parity

Specifies whether redundancy is provided by using parity.

stripe

Specifies how many columns a volume should have.

striped

Specifies whether a volume is striped.

For example, the following rule specifies that a volume can be created using parity to provide data redundancy:


parity true

Volume Template

A volume template (or template for short) is a meaningful collection of rules that provide a capability. A template can specify one or more capabilities that a volume created using the template may have, and consists of a collection of storage selection and layout rules. For example, a template may allow you to create a volume that is able to tolerate the failure of a certain number of controllers, or that has a certain number of copies of the volume data.

When creating a volume, it is easier to specify its desired capabilities than to specify the precise layout of the volume on the available storage. ISP selects the appropriate templates and uses them to create a volume with the desired capabilities.

If you specify parameter values for a volume's capabilities, the rules that are defined within the chosen template use these values when selecting and laying out storage. If not specified, the default parameter values for a volume's capabilities are assumed by the template.

The following table shows some simple examples of templates and the capabilities that they might provide:

Template Provides Capabilities Adjustable Parameters for the Capability

DataMirroring

DataMirroring

nmirs --- number of mirrors (plexes)

Raid5Volume

Raid5Capability, Raid5LogMirroring

ncols --- minimum number of columns

nlogs --- number of log copies

nmaxcols --- maximum number of columns

Striping

Striping

ncols --- minimum number of columns

nmaxcols --- maximum number of columns

See Volume Templates for details of the templates that are provided.

Template Sets

A template set consists of related capabilities and templates that have been collected together for convenience. Associating a template set with a storage pool is equivalent to associating each of its member templates separately with the storage pool. See Template Sets for details of the template sets that are provided.

User Template

A user template (or user-defined template) defines an arbitrary collection of capabilities, templates and rules to which you want volumes of a certain type to conform. For example, you might want all volumes that you create to store database tables to share the same reliability and performance capabilities, and also that they only be allocated from a restricted set of storage.

It is useful to create user templates if you regularly create volumes with similar capabilities.

The following table shows some examples of user templates that might be created:

User Template Description

OracleTable

Provides a reliable high-performance volume that is suitable for use by a database table.

OracleIndex

Provides an extremely high-performance volume that is suitable for a database index.

As for templates, each user template can have a number of adjustable parameters that you can use for tuning the characteristics of the created volumes.


Note   Note    A user template is not directly associated with a storage pool. Its association is implied by its reference to capabilities and templates.

Application Volume

An application volume is created by ISP, and then exported for use by an application such as a database or file system. It is identical to a traditional volume that you would create using vxassist and other VxVM commands, except that you can only administer it by using the vxvoladm command, or via the interface to ISP that is provided by the VEA. Similarly, traditional volumes cannot be administered by using the vxvoladm command, or via the VEA interface to ISP.


Note   Note    In this book, the term volume always means application volume unless it is specified that a traditional VxVM volume is meant.

Intent

The intent of a volume is a conceptualization of its purpose as defined by its characteristics and implemented by a template. ISP attempts to preserve the intent of a volume whenever the volume is reconfigured, resized, or relocated. Intent preservation automatically conserves capabilities such as reliability and performance, and observes additional rules such as allocating storage based on confinement and exclusion specifications.

 ^ Return to Top Previous  |  Next  >  
Product: Storage Foundation Guides  
Manual: Storage Foundation 4.1 Intelligent Storage Provisioning Administrator's Guide  
VERITAS Software Corporation
www.veritas.com