When you create a remote volume mirror, a mirrored volume pair is created and consists of a primary volume at a primary storage array and a secondary volume at a secondary storage array. A stan dar d volume may only be included in one mirrored volume pair. There can be up to 8 defined mirrors on a SGI TP9400 storage array and up to 32 defined mirrors on a SGI TP9500 storage array.

The primary volume is the volume that accepts host I/O and stores application data. When the mirror relationship is first created, data from the primary volume is copied in its entirety to the secondary volume. This process is known as a full synchronization and is directed by the controller owner of the primary volume. During a full synchronization, the primary volu me remains fully ac cessible for all normal I/O operations.

The controller owner of the primary volume initiates remote writes to the secondary volume to keep the data on the two volumes synchronized. Whenever the data on the primary volume and the secondary volume becomes unsynchronized, the controller owner of the primary volume initiates a full synchronization.

The secondary volume maintains a mirror (or copy) of the data on its associated primary volume. The controller ow ner of the secondary volume receives remote writes from the primary volume controller owner, but will not accept host read or write requests.

The secon dary volume remains unavailable to host applications while mirroring is underway. In the event of a disaster or catastrophic failure of the primary site, a role reversal can be performed to promote the secondary volume to a primary role. Hosts will then be able to access the newly promoted volume and business operations can continue.

Figure 3-1 shows the primary and secondary volumes displayed in the Array Management window for the primary storage array and the secondary storage array.

A mirror repository volume is a special volume in the storage array created as a resource for the controller owner of the primary volume in a Remote Volume Mirror. The controller stores mirroring information on this volume, including information about remote writes that are not yet complete. The controller can use this information to recover from contr oller resets and accidental powering-down of storage arrays.

When you activate the Remote Volume Mirroring premium feature on the storage array, you create two mirror repository volumes, one for each controller in the storage array. An individual mirror repository volume is not needed for each Remote Volume Mirror.

When you create the mirror repository volumes, you specify the location of the volumes. You can either use existing free capacity or you can create a volume group for the volumes from unconfigured capacity and then specify the RAID level.

Figure 3-2 shows the mirror repository volumes displayed in the Array Management window for the primary storage array.

Because of the critical nature of the data being stored, the RA ID level of mirror repository volumes must not be RAID 0 (for data striping). The required size of each vol ume is 4 MB, or 8 MB total for both mirror repository volumes of a dual controller array.

Prior to creating a mirror relati onship, the Remote Volume Mirroring premium feature must be enabled on both the primary and secondary storage arrays. A secondary volume candidate must be created on the secondary site if one does not already exist and must be a standard volume of equal or greater capacity than the associated primary volume.

When secondary vol ume candidates are available, a mirror relationship can be established in the storage management software by identifying the storage array containing the primary volume and the storage array containing the secondary volume.

When the mirror relationship is first created, a full synchronization automatically occurs, with data from the primary volume copied in its entirety to the secondary volume. For more information on creating mirror relationships, see “Creating Mirror Relationships”.

When processing write requests, the primary controller may be able to write to the primary volume, but a link interruption prevents communication with the remote secondary controller.

In this case, the remote write cannot complete to the secondary volume and the primary and secondary volumes are no longer correctly mirrored. The primary controller transitions the mirrored pair into an Unsynchronized status and sends an I/O completion to the primary host, as shown in Figure 3-4. The primary host can continue to write to the primary volume but remote writes will not take place.

When connectivity is restored between the controller owner of the primary volume and the controller owner of the secondary volume, a full synchronization will automatically take place. The mirrored pair will then transition from an Unsynchronized status to a Synchronization-in-Progress status, as shown in Figure 3-5.

The primary controller will also mark the mirrored pair as Unsynchronized when a volume error on the secondary side prevents the remote write from completing. For example, an offline or a failed secondary volume can cause the Remote Volume Mirror to become Unsynchronized. When the volume error is corrected (the secondary volume is placed online or recovered to an Optimal status), then a full synchronization automatically begins, and the mirrored pair transitions to a Synchronization-in-Progress status.

For more information on Remote Volume Mirroring statuses, see “Using the Remote Volume Mirroring Premium Feature”.

Three Fibre Channel fabric configurations are supported in this version of the storage management software. The level of redundancy available will depend on the type of configuration selected.

For instructions on how to configure storage arrays, see “Hardware and Software Installation”.

The controller owner of the primary volume will only attempt to communicate with its mirrored controller in the secondary storage array. As shown in Figure 3-6, Controller A in the primary storage array only attempts communication with Controller A in the secondary storage array.

The controller (A or B) that owns the primary volume determines the controller owner of the secondary volume. If the primary volume is owned by Controller A on the primary storage array, the secondary volume is therefore owned by Controller A on the secondary storage array. If primary Controller A cannot communicate with secondary Controller A, no controller ownership changes take place.

When an I/O path error causes a volume ownership change on the primary storage array, or if the storage administrator changes the controller owner of the primary volume, the next remote write processed will automatically trigger an ownership change on the secondary storage array.

As shown in Figure 3-7, if a primary volume is owned by Controller A and the controller owner is changed to Controller B, the next remote write will change the con troller owner of the secondary volume from Controller A to Controller B.

Because controller ownership changes on the secondary storage array are controlled by the primary controller, they do not require any special intervention and cannot be manually changed by the storage administrator.

The maximum distance permitted between storage arrays participating in a mirror relationship is governed by the distance limits of the Fibre Channel links. Using standard Single-Mode Fiber technology, a maximum link distance of 10 km between storage arrays can be achieved.

The following general performance considerations should be noted when creating Remote Volume Mirrors.

Storage Partitioning is a premium feature that enables hosts to share access to volumes in a storage array. A storage partition is created when you define a collection of hosts (a host group) or a single host and then define a volume-to-logical unit number (LUN) mapping. This mapping allows you to define what host group or host will have access to a particular volume in the storage array.

Figure 3-8 shows the Mappings View of the Array Management window, where Primary Volume Accounting has been included in a partition accessible by Host Group Campus East, and Secondary Volume Accounting 2 has been included in a partition accessible by Host Group Campus West.

The storage partition definitions for the primary and secondary storage arrays are independent of each other. If these def initions are put in place while the volume is in a secondary role, it will reduce the administrative effort associated with site recovery if it becomes necessary to promote the volume to a primary role. For more information on defining storage partitions, see the Array Management window online help.

A snapcopy volume is a point-in-time image of a volume and is typically created so that an application, such as a backup application, can access the snapcopy volume and read the data while the base volume remains online and accessible to hosts.

The volume for which the point-in-time image is created is known as the base volume and must be a standard volume in the storage array. For the Remote Volume Mirroring feature, the base volume of a snapcopy volume is permitted to be a candidate for the primary volume in a mirror relationship. The snapcopy repository volume stores information about all data altered since the snapcopy was created. Figure 3-9 shows the primary volume (also the base volume for the snapcopy volume), the secondary volume, and the snapcopy volume in the Array Management window for the primary storage array.

Secondary v olumes are restricted from being the base volume for a snapcopy volume because any full synchronization process would fill the snapcopy repository volume, resulting in a failed snapcopy volume.

Snapcopy Volume Failure

In this version of the storage management software, snapcopy volumes based on the primary volume of a mirrored volume pair are permitted.

If a role reversal is performed and the primary volume is demoted to the secondary volume role, all snapcopy volumes associated with the demoted primary volume will be failed.

As shown in Figure 3-10, Primary Volume Accounting and Secondary Volume Accounting 2 are displayed in the Logical view of the Array Management window for the primary storage array. A snapcopy volume of Primary Volume Accounting has been created.

If a role reversal is performed, Primary Volume Accounting is demoted to a secondary role and the snapcopy volume fails, as shown in Figure 3-11. This is because the Secondary Volume Accounting 2 is now considered as the base volume of the snapcopy volume, which is not permitted.

If a snapcopy volume fails as the result of a role reversal, you will need to perform another role reversal to promote the secondary volume back to a primary role. After the role reversal is complete, the snapcopy volume will return to an Optimal status. For more information, see “Troubleshooting”.

The following are the minimum hardw are requirements needed when configuring your system to use the Remote Volume Mirroring premium feature. The list includes components that are supplementary to the standard required for a basic storage array environment as described in the SGI TP9500 RAID Installation and Upgrade Guide.

The following are the minimum software requirements needed when configuring your system to use the Remote Volume Mirroring premium feature.

Certified Host Operating Systems

Hosts accessing mirrored volumes should be running one of the following certified operating systems when using the Remote Volume Mirroring feature:

The Highest Availability Campus Configuration is a fully redundant configuration, and consists of two storage arrays and four switches connected using a Fibre Channel fabric as shown in Figure 3-12. The primary storage array and secondary storage array may have a maximum connection distance of up to 10 km (or 6.25 miles).

For detailed instructions on setting up this configuration, see “Hardware and Software Installation”.

The Campus Configuration is a lower-cost configuration and consists of two storage arrays and two switches, connected using a Fibre Channel fabric as shown in Figure 3-13. The primary storage array and secondary storage array may have a maximum connection distance of up to 10 km (or 6.25 miles).

For detailed instructions on setting up this configuration, see “Hardware and Software Installation”.

The Intra-Site Configur ation is the lowest-cost configuration which consists of two storage arrays and two switches, connected using a Fibre Channel fabric as shown in Figure 3-14. The primary storage array and secondary storage array may have a maximum connection distance of up to 500 m (or 0.32 miles) and would typically be located in the same building or in the same room.

For detailed instructions on setting up this configuration, see “Hardware and Software Installation”.

Complete the tasks in Table 3-1 before starting the storage management software.

Complete the tasks in Table 3-2 before establishing mirror relationships in your storage arrays.

The Remote Volume Mirroring feature utilizes Fibre Channel switches creating a fabric environment for data replication. These switches require only minimal additional site preparation requirements beyond basic storage array operation.

For basic site preparation considerations including the use of Fibre Channel cables and connections, see the SGI TP9400 and SGI TP9500 RAID Owner's Guide.

For additional site preparation considerations concerning the Fibre Channel switches, including power requirements and physical dimensions/requirements, see the documentation provided by the switch manufacturer.

Ensure that all of the required hardware and software are available before proceeding. For a list of requirements, see “Hardware and Software Installation”.

Because of possible restrictions at the host level, the supported Remote Volume Mirroring configurations contain Fibre Channel switches zoned so a single host adapter can access only one controller per storage array . Additionally, all configurations utilize a separate zone for the ports reserved for the Remote Volume Mirroring feature.

Switch zoning configurations are typically set up using management software provided by the manufacturer of the Fibre Channel switch. This software should be included with the materials provided when the switch was purchased.

When two or more Fibre Channel switches are cascaded together, the switch management software will combine the ports for all switches that are linked. For example, if two 16-port Fibre Channel switches are cascaded with a physical connection using a Fibre Channel cable, the switch management software will show ports 0 through 31 participating in the fabric rather than 2 switches each with ports 0 through 15. Therefore, a zone that is created containing any of these ports can exist on multiple cascaded switches.

In Figure 3-15, the top set of switches are on the same network and cascaded, therefore Zone 1 is the same zone on Switch 1A as Zone 1 on Switch 1B. The bottom set of switches are on the same network, but are not cascaded. Even though both contain a Zone 1 (shown as Zone A in Switch 2), these zones are independent of each other.

For more information about Fibre Channel switch zoning or setting up a zone configuration, see the manufacturer's documentation provided with the switch.

Due to the varying Remote Volume Mirroring configurations, the switch zone settings are presented preceeding each configuration illustration in this chapter.

This configuration has two Fibre Channel switches at both the primary and secondary sites to provide for complete failover and redundancy for failures in Fibre Channel switches and cables, in addition to any host or storage array failure. Two Fibre Channel switches at each site also provide redundancy to the local site in addition to a fully redundant remote configuration. There is no single point of failure in the hardware components.

Figure 3-16 shows the Highest Availability Campus Configuration after installation is complete.

Switch Zoning for the Highest Availability Campus Configuration

This configuration allows for a separate zone for each reserved p ort for the Remote Volume Mirroring feature.

The switches do not need to be zoned exactly as presented in this configuration. However, the following are requirements when zoning switches for the Highest Availability Campus Configuration.

The switches in Figure 3-17 contain 16 ports each, which leaves many unused ports per switch. The remaining ports can be distributed among the other zones. However, it is recommended that most remaining ports be assigned to the zones containing the host connections (Zones 1 and 2 in Figure 3-17). This will allow easy set up for additional hosts to connect to the environment.

Figure 3-17 shows how the four switches are zoned for the Highest Availability Campus Configuration.

Given the requirements and the zoning shown in Figure 3-17, ensure that all four switches are properly zoned before proceeding. For more information on Zoning switches, see “Switch Zoning Overview”.

Cabling for the Highest Availability Campus Configuration

Table 3-3 is provided for a convenient reference of port connections.

After the four Fibre Channel Switches are properly zoned, complete the following procedure to set up the Highest Availability Campus Configuration for the Remote Volume Mirroring feature.

This configuration offers the same functionality as the Highest Availability Campus Configuration, but contains only one switch at each site, rather than two. The configuration is still redundant for host bus adapters, controllers, and Remote Volume Mirroring ports, but is a single point of failure for switches. If a switch at either site fails, the Remote Volume Mirroring Feature cannot function. For this reason, the Highest Availability Campus Configuration is highly recommended for total environment redundancy.

Figure 3-21 shows the Campus Configuration after installation is complete.

Switch Zoning for the Campus Configuration

This configuration allows for a separate zone for each reserved port for the Remote Volume Mirroring feature.

The switches do not need to be zoned exactly as presented in this configuration. However, the following are requirements when zoning switches for the Campus Configuration.

The switches in Figure 3-22 contain 16 ports each, which leaves many unused ports per switch. The remaining ports can be distrib uted among the other zones. However, it is recommended that most remaining ports be assigned to the zones containing the host connections (Zone 1 in Figure 3-22). This will allow easy set up for additional hosts to connect to the environment.

Figure 3-22 shows how the two switches are zoned for the Campus Configuration.

Given the requirements and the zoning example shown in Figure 3-22, ensure that both switches are properly zoned before proceeding. For more information on Zoning switches, see “Switch Zoning Overview”.

Cabling for the Campus Configuration

Table 3-4 is provided for a convenient reference of port connections.

After both Fibre Channel Switches are properly zoned, complete the following procedure to set up the Campus Configuration for the Remote Volume Mirroring feature.

This configuration is u sed in environments where a long distance fabric is not required because of the close proximity of the hosts and storage arrays. The configuration is still redundant for host bus adapters, controllers, Remote Volume Mirroring ports, and switches, but is a single point of failure for the site since all equipment can be destroyed by the same disaster. For this reason, the Highest Availability Campus Configuration is highly recommended for total environment redundancy.

Figure 3-26 shows the Intra-Site Configuration after installation is complete.

Switch Zoning for the Intra-Site Configuration

This configuration is designed for switch redundancy. However, the switches are not cascaded, and are therefore independent of each other. The switches do not need to be zoned exactly as presented in this configuration. However, the following are requirements when zoning switches for the Intra-Site Configuration.

The switches in Figure 3-27 contain 16 ports each, which leaves many unused ports per switch. The remaining ports can be distribut ed among the other zones. However, it is recommended that most remaining ports be assigned to the zones containing the host connections (Zone 1 and Zone A in Figure 3-27). This will allow easy set up for additional hosts to connect to the environment.

In this example, the switches use half of the ports for each zone for simplicity, although Zone 2 and Zone B require less ports.

Given the requirements and the zoning example shown in Figure 3-27, ensure both switches are properly zoned before proceeding. For more information on zoning switches, see “Switch Zoning Overview”.

Cabling for the Intra-Site Configuration

Table 3-5 is provided for a convenient reference of port connections.

After both Fibre Channel Switches are properly zoned, complete the following procedure to set up the Intra-Site Configuration for the Remote Volume Mirroring feature.

All hardware and software is n ow set up to support mirror relationships through the Remote Volume Mirroring feature. However, before creating mirror relationships, ensure that all required components are working properly.

Complete the following procedure for each storage array that will participate in any aspect of a mirror relationship to verify correct configuration.

To begin working with Mirror Relationships, go to “Using the Remote Volume Mirroring Premium Feature”.

There are four possible statuses of the Remote Volume Mirroring feature: Disabled/Deactivated, Disabled/Activated, Enabled/Deactivated, and Enabled/Activated.

The current status can be determined by hovering the mouse pointer over the Remote Volume Mirroring feature icon in the premium feature status area of the Array Management window, as shown in Figure 3-32.

The status of the Remote Volume Mirroring feature is one of the following.

If the current status of the Remote Volume Mirroring feature is Disabled/Deactivated or Disabled/Activated, complete the following to enable the Remote Volume Mirroring feature.

Activating the Remote Volume Mirroring feature prepares the storage array to create and configure mirror relationships. After the feature is activated, the secondary ports for each controller are reserved and de dicated to Remote Volume Mirroring feature usage. In addition, a mirror repository volume is automatically created for each controller in the storage array.

If the current status of the Remote Volume Mirroring feature is Enabled/Deactivated, complete the following to activate the premium feature.

Before any mirror relationships can be created, volumes must exist at both the primary and secondary sites. The volume residing in the primary storage array is the primary volume. Similarly, the volume residing in the secondary storage array is the secondary volume.

If a primary volume does not exist, one will need to be created on the primary storage array.

If a secondary volume does not exist, one will need to be created on the secondary storage array.

Consider the following when creating the secondary volume:

For more information about creating volumes, see the Array Management window online help.

When adequate volumes exist at both sites, mirror relationships can be created using the Remote Volume Mirror Wizard.

The Create Remote Volume Mirror Wizard allows a user to create a mirror relationship between a primary volume and a secondary volume.

The mirror relationship that exists between the primary volume and the secondary volume can be examined by various methods, including: the Storage Array Profile, the Mirroring Properties dialog, and the View Associated Components dialog.

The Storage Array Profile

The Storage Array Profile is the most efficient way to view information about any or all components of the storage array. Details for all volumes, including the primary, secondary, and mirror repository volumes can all be easily viewed through the Storage Array Profile simultaneously. The Storage Array Profile also contains specific information for components associated with mirror relationships.

Complete the following procedure to display a Storage Array Profile.

The Mirroring Properties Dialog

The Mirroring Properties dialog displays all the physical characteristics of a single volume in the mirror relationship. The dialog displays the same information as the Storage Array Profile for the selected volume, but is specific to that volume. In addition, the synchronization progress is displayed if the selected volume is synchronizing data with another volume in a mirror relationship.

Complete the following procedure for each volume that needs to be examined in the mirror relationship.

The View Associated Components Dialog

The View Associated Components dialog provides a more graphical representation of the volumes participating in the mirror relationship. In addition, more detail is given for all components, rather than only the selected volume.

Complete the following procedure to view all associated components in a mirror relationship, including primary, secondary, and mirror repository volumes.

The Synchronization Priority Level of a mirror relationship defines the amount of system resources used to synchronize the data between the primary and secondary volumes of a mirror relationship. If the highest priority level is selected for a mirror relationship, the data synchronization uses a high amount of system resources to perform the full synchronization, but may decrease performance for host I/O, including other mirror relationships. Conversely, if the lowest synchronization level is selected, there is less impact on complete system performance, but the full synchronization may be slower.

To change the Synchronization Priority Level for a mirror relationship, complete the following procedure.

Removing a mirror relationship between a primary and secondary volume does not affect any of the existing data on either volume. The link between the volumes is removed, but the primary volume still continues normal I/O operation. This action can be performed for backup routines, particularly from the secondary volume. A mirror relationship between the two volumes can be re-created unless one of the volumes is deleted.

To remove a mirror relationship between two volumes, complete the following procedure.

Deleting a volume participating in a mirror relationship removes the mirror relationship and completely deletes the volume from the storage array. The mirror relationship cannot be redefined until a new volume is created or an alternate volume is chosen to replace the deleted volume.

To delete a primary or secondary volume from a mirror relationship, complete the following procedure.

To change the Remote Volume Mirroring feature status from Enabled/Activated to Enabled/Deactivated or from Disabled/Activated to Disabled/Deactivated, complete the following procedure.

To change the Remote Volume Mirroring feature status from Enabled/Deactivated to Disabled/Deactivated or from Enabled/Activated to Disabled/Activated, complete the following procedure.

Backing up critical data regularly is vital insurance against disasters or catastrophic failure. Backups should be performed regardless of whether the Remote Volume Mirroring feature is in use.

A backup of all critical data on both the primary and secondary storage arrays participating in Remote Volume Mirroring should be performed prior to creating a Remote Volume Mirror. Periodic backups of the primary storage array should be carried out after Remote Volume Mirrors are in use.

Backups to tape or disk can be made while I/O continues using the Snapcopy Volume premium feature, which captures a point-in-time image of a volume. For more information on using snapcopy volumes, see “Snapcopy Volumes” or the Array Management window online help.

When preparing for a potential disaster, ensure the secondary storage array, to which the data is being mirrored, is structured in the same manner as the primary storage array.

Prior to creating a Remote Volume Mirror, you will need to prepare your secondary storage array to ensure that storage partitions are defined and ready to function if secondary storage array volumes are promoted to primary roles.

Defining Storage Partitioning

The Storage Partitioning premium feature allows you to define which hosts have access to which volumes on the storage array and at what logical unit number. When configuring your secondary storage array, replicate the storage partitions that were previously defined on the primary storage array. This will ensure that after a role reversal is carried out, hosts will be able to access their required data.

For more information on Storage Partitioning, see “Using Other Premium Features and Remote Volume Mirroring” or the Array Management window online help.

Host failover clustering is a process in which the operating system and application software work together to provide continuous availability in the event of an application failure, hardware failure, or operating-system error. Software products that support remote failover may refer to global or geographic cluster management.

The Remote Volume Mirroring feature can be used in conjunction with remote host failover to provide complete redundancy of host systems and storage arrays. If a disaster occurs at the primary storage array, hosts at the secondary storage array will take over processing and initiate a role change to the secondary storage array.

Remote failover is not a requirement for a disaster recovery plan. Depending on the requirements for how quickly operations must be functioning at Optimal status again, mirroring the storage arrays and depending on manual startup of the remote hosts may be acceptable.

The checklist in Table 3-6 should be completed in preparation for disaster recovery. Complete the steps listed to ensure access to mirrored data is possible in the event of a disaster or catastrophic failure.

Entire site failover is necessary when an event occurs that inhibits normal operation to the primary site for an extended period of time.

If a disaster occurs to the primary or secondary site that renders the storage array or all data on the storage array irreparable, then the site needs to be rebuilt with new equipment.

Complete the following procedure to rebuild a damaged site:

After the damaged site is back online and properly configured, mirror relationships can be resumed by completing the following procedure.

A secondary volume is typically promoted to a primary volume role when a catastrophic failure has occurred to the storage array that contains the primary volume, and the secondary volume needs to be promoted so that hosts can access data and business operations can continue.

This option is not available unless the Remote Volume Mirroring premium feature has been activated.

If a communication problem between the secondary and primary sites prevents the demotion of the remote primary volume, an error message is displayed. However, you are given the opportunity to proceed with the promotion of the secondary volume, even though this will lead to a dual-primary condition.

To promote a secondary volume to a primary role:

After the controller owner of the primary volume can be contacted, the primary volume is automatically demoted back to be the secondary role in the mirror relationship.

Changing a primary volume to a secondary role is used for role reversals during normal operating conditions. You can also use this option during a Recovery Guru procedure when a dual-primary condition occurs.

This option is not available unless the Remote Volume Mirroring premium feature has been activated.

If a communication problem between the primary and secondary sites prevents the promotion of the secondary volume, an error message is displayed. However, you are given the opportunity to proceed with the demotion of the primary volume, even though this will lead to a dual-secondary condition.

If the selected primary volume has associated snapcopy volumes, demoting this volume to a secondary role will cause the associated snapcopy volumes to fail.

To demote a primary volume to a secondary role:

Once the controller owner of the secondary volume can be contacted, the secondary volume is automatically promoted back to the primary volume in the mirror relationship.