5.3 HPE MSA Data Protection & Management: Snapshots, Volume Copy, Remote Snap & SMU v4

HPE MSA Data Protection & Management

Quick Answer: Every MSA Gen7 includes 64 snapshots. The Advanced Data Services (ADS) Suite (included with the MSA 2072, optional on the MSA 2070) raises this to 512 snapshots and adds Volume Copy and Remote Snap, the MSA's asynchronous, snapshot-based replication over FC or iSCSI. Manage the array with the Storage Management Utility (SMU) v4, the CLI or REST APIs, and use the MSA Health Check and online firmware updates to keep it healthy.

Snapshots

MSA snapshots are point-in-time copies of virtual volumes:

• They are created instantly and consume space only as data changes, because MSA virtual snapshots use a redirect-on-write design: new writes go to new pages while the snapshot keeps pointing to the original pages.
• A snapshot can be mapped to a host like a volume, for example to restore files, run a backup off-host or test changes.
• Recovery is instant: the volume can be reverted (reset) to a previous snapshot.
• Snapshots of snapshots form a snapshot tree.
• Snapshot space management lets the administrator limit how much of the pool snapshots may use and choose what happens at the limit, such as notification or deletion of older snapshots.

HPE also notes that snapped data can be backed up to disk, virtual tape or physical tape without a backup window, because the backup reads the snapshot rather than the live volume.

Volume Copy

A Volume Copy creates a complete, independent copy of a volume. HPE lists up to 128 volume copies. When the copy completes, it becomes a standard volume with no dependency on the source, which suits:

• A test or development copy that must not share pages with production.
• Moving a volume to the other pool, for example to rebalance load between controllers.

Remote Snap replication

Remote Snap is the MSA's array-based asynchronous replication. It copies block data from a volume on one MSA to a volume on a second, independent MSA, which may be in the same room or at a remote site.

How a replication cycle works:

1. A snapshot of the primary volume freezes a point-in-time image.
2. It is compared with the snapshot from the previous replication.
3. Only the changed data is sent over TCP/IP (iSCSI) or Fibre Channel to the destination volume.
4. The destination keeps its own snapshots, which give multiple recovery points.

Key facts for the exam:

• Remote Snap works between all MSA Gen7 models that use virtual storage.
• It supports many-to-1 replication of up to 4 nodes, typically several branch arrays replicating to a central array.
• It supports failover to the secondary volume or volume group, mapping the secondary for access, failback with the option to keep changes made at the secondary, and reversing the replication direction.
• Licensing: one ADS license per array. Two arrays replicating to each other need two licenses, even if one of them is an MSA 2072 that already includes its license.
• Because it is asynchronous, the RPO equals the replication interval plus transfer time; it is not zero.

Encryption and integrations

Management interfaces

Each controller has an out-of-band Ethernet management port. Because both controllers serve SMU, losing one controller's management connection does not lose management of the array. LDAP integration lets administrators log in with Active Directory accounts in manage or monitor roles.

Health, alerts and firmware

• MSA Health Check: reached from SMU v4 or HPE.com, it analyzes array information and reports problems and best-practice issues. HPE added simpler and faster access to it in Gen7.
• Alerts: email (SMTP) and SNMP notifications for critical, error and warning events.
• Firmware Update Server: automated notification that new firmware is available.
• Online firmware updates: Gen7 updates controller and drive firmware non-disruptively through MSA Smart Components. This requires multipathing on the hosts so that I/O can move to one controller while the other updates.
• Partner Firmware Update (PFU): keeps both controllers on the same firmware by updating the partner automatically.

Scenario: branch offices to headquarters

A retailer has three branch MSA 2070 arrays and one MSA 2072 at headquarters. Each branch replicates its file-server and point-of-sale volumes every hour to headquarters with Remote Snap over the WAN using iSCSI. This is a many-to-1 design of four arrays in total, which is the documented maximum. All four arrays need an ADS license: the three 2070s need purchased licenses, and the 2072 uses its included license. After a branch failure, the administrator fails over to the replica at headquarters, maps it to a recovery host, and later fails back with the changes.

Common exam traps

• Remote Snap is asynchronous only. It cannot give RPO = 0.
• Licenses are per array, on both the source and the destination.
• Snapshots are not backups by themselves. They live in the same pool; combine them with Remote Snap, backup or tape.
• Online firmware updates need host multipathing.