Storage virtualization terminology

Although storage virtualization is a term that is used extensively throughout the storage
industry, it can be applied to a wide range of technologies and underlying capabilities. In
reality, most storage devices can technically claim to be virtualized in one form or another.

Tekmart Support Team

Therefore, we must start by defining the concept of storage virtualization as used in this tutorial.

IBM interprets storage virtualization in the following manner:
Storage virtualization is a technology that makes one set of resources look and feel like another set of resources, preferably with more desirable characteristics.
It is a logical representation of resources that are not constrained by physical limitations:
– It hides part of the complexity.
– It adds or integrates new functionality with existing services.
– It can be nested or applied to multiple layers of a system.
When considering storage virtualization, it is important to understand that virtualization can be implemented at various layers within the input/output (I/O) stack. We must clearly distinguish between virtualization at the disk layer and virtualization at the file system layer.

The Storage Networking Industry Association’s (SNIA) block aggregation model provides a useful overview of the storage domain and its layers. The figure shows the three layers of a storage domain:

• The file
• The block aggregation
• The block subsystem layers

The model splits the block aggregation layer into three sublayers. Block aggregation can be realized within hosts (servers), in the storage network (storage routers and storage controllers), or in storage devices (intelligent disk arrays).
One of the IBM implementations of a block aggregation solution is IBM Storwize V7000 Gen2. Storwize V7000 Gen2 is implemented as a clustered appliance in the storage network layer.

The Storwize V7000 Gen2 solution provides a modular storage system that includes the capability to virtualize both external SAN-attached storage and its own internal storage. The Storwize V7000 Gen2 solution is built upon the IBM SAN Volume Controller technology base, and uses technology from the IBM System Storage DS8000® family.

A Storwize V7000 Gen2 system provides several configuration options that are aimed at simplifying the implementation process. It also provides automated wizards, called Directed Maintenance Procedures (DMP), to help resolve any events that might occur. A Storwize V7000 Gen2 system is a midrange, clustered, scalable, and external virtualization device.

Included with a Storwize V7000 Gen2 system is a graphical user interface (GUI) that enables storage to be deployed quickly and efficiently. The GUI runs on the Storwize V7000 Gen2 system, so there is no need for a separate console. The management GUI contains a series of preestablished configuration options that are called presets, and that use common settings
to quickly configure objects on the system. Presets are available for creating volumes and FlashCopy mappings, and for setting up a RAID configuration.

The Storwize V7000 Gen2 solution provides a choice of up to 1056 serial-attached Small Computer System Interface (SCSI), called SAS, drives for the internal storage in a clustered system. It uses SAS cables and connectors to attach to the optional expansion enclosures. In a clustered system, the Storwize V7000 Gen2 can provide about 4 pebibytes (PiB) of internal
raw capacity.

When virtualizing external storage arrays, a Storwize V7000 Gen2 system can provide up to 32 PiB of usable capacity. A Storwize V7000 Gen2 system supports a range of external disk systems, similar to what the IBM SAN Volume Controller supports today.

The Storwize V7000 Gen2 solution consists of 1 – 4 control enclosures and, optionally, up to 80 expansion enclosures. It also supports the intermixing of the different expansion enclosures. Within each enclosure are two canisters. Control enclosures contain two node canisters, and expansion enclosures contain two expansion canisters.

RAID
The Storwize V7000 Gen2 setup contains several internal drive objects, but these drives cannot be directly added to storage pools. The drives need to be included in a RAID to provide protection against the failure of individual drives.
These drives are referred to as members of the array. Each array has a RAID level. RAID levels provide various degrees of redundancy and performance, and have various restrictions regarding the number of members in the array.
Storwize V7000 Gen2 supports hot spare drives. When an array member drive fails, the system automatically replaces the failed member with a hot spare drive and rebuilds the array to restore its redundancy. Candidate and spare drives can be manually exchanged with array members.

Each array has a set of goals that describe the location and performance of each array. A sequence of drive failures and hot spare takeovers can leave an array unbalanced (with members that do not match these goals). The system automatically rebalances such arrays when the appropriate drives are available.

The following RAID levels are available:
RAID 0 (striping, no redundancy)
RAID 0 arrays stripe data across the drives. The system supports RAID 0 arrays with just one member, which is similar to a traditional just a bunch of disks (JBOD) unconfigured attach. RAID 0 arrays have no redundancy, so they do not support hot spare takeover or immediate exchange. A RAID 0 array can be formed by 1 – 8 drives.

RAID 1 (mirroring between two drives)
RAID 1 arrays stripe data over mirrored pairs of drives. A RAID 1 array mirrored pair is rebuilt independently. A RAID 1 array can be formed by two drives only.

RAID 5 (striping, can survive one drive fault)
RAID 5 arrays stripe data over the member drives with one parity strip on every stripe. RAID 5 arrays have single redundancy. The parity algorithm means that an array can tolerate no more than one member drive failure. A RAID 5 array can be formed by 3 – 16 drives.

RAID 6 (striping, can survive two drive faults)
RAID 6 arrays stripe data over the member drives with two parity stripes (known as the P-parity and the Q-parity) on every stripe. The two parity strips are calculated using different algorithms, which give the array double redundancy. A RAID 6 array can be formed by 5 to 16 drives.

RAID 10 (RAID 0 on top of RAID 1)
RAID 10 arrays have single redundancy. Although they can tolerate one failure from every mirrored pair, they cannot tolerate two-disk failures. One member out of every pair can be rebuilding or missing at the same time. A RAID 10 array can be formed by 2 – 16 drives.

MDisks
A managed disk (MDisk) is the unit of storage that Storwize V7000 Gen2 virtualizes. This unit could be a logical volume on an external storage array presented to Storwize V7000 Gen2, or a RAID array consisting of internal drives. Storwize V7000 Gen2 can then allocate these MDisks into various storage pools. An MDisk is not visible to a host system on the SAN,
because it is internal or zoned only to the Storwize V7000 Gen2 system.

The MDisks are placed into storage pools where they are divided into several extents, which can range in size from 16 megabytes (MB) – 8182 MB, as defined by the storage administrator.

A volume is host-accessible storage that has been provisioned out of one storage pool or, if it is a mirrored volume, out of two storage pools. The maximum size of an MDisk is 1 PB. A Storwize V7000 Gen2 system supports up to 4096 MDisks (including internal RAID arrays).

At any point in time, an MDisk is in one of the following four modes:

Array
Array mode MDisks are constructed from drives using the RAID function. Array MDisks are always associated with storage pools.

Unmanaged MDisk
An MDisk is reported as unmanaged when it is not a member of any storage pool. An unmanaged MDisk is not associated with any volumes, and has no metadata stored on it. Storwize V7000 Gen2 does not write to an MDisk that is in unmanaged mode, except when it attempts to change the mode of the MDisk to one of the other modes.

Storwize V7000 Gen2 can see the resource, but the resource is not assigned to a storage pool.

Managed MDisk
Managed mode MDisks are always members of a storage pool, and they contribute extents to the storage pool. Volumes (if not operated in image mode) are created from these extents. MDisks operating in managed mode might have metadata extents allocated from them, and can be used as quorum disks. This mode is the most common and normal mode for an MDisk.

Image mode MDisk
Image mode provides a direct block-for-block translation from the MDisk to the volume by using virtualization. This mode is provided to satisfy three major usage scenarios:

• – Image mode enables the virtualization of MDisks already containing data that was written directly, and not through Storwize V7000 Gen2. Rather, it was created by a direct-connected host. This mode enables a client to insert Storwize V7000 Gen2 into the data path of an existing storage volume or LUN with minimal downtime. The image mode is typically used for data migration from old storage systems to new.
• – Image mode enables a volume that is managed by Storwize V7000 Gen2 to be used with the native copy services function provided by the underlying RAID controller. To avoid the loss of data integrity when Storwize V7000 Gen2 is used in this way, it is important that you disable the Storwize V7000 Gen2 cache for the volume.
• – Storwize V7000 Gen2 provides the ability to migrate to image mode, which enables Storwize V7000 Gen2 to export volumes and access them directly from a host without the Storwize V7000 Gen2 in the path.

Each MDisk presented from an external disk controller has an online path count that is the number of nodes having access to that MDisk. The maximum count is the maximum number of paths detected at any point in time by the system. The current count is what the system sees currently. A current value less than the maximum can indicate that SAN fabric
paths have been lost.

SSDs (flash drives) that are in Storwize V7000 Gen2 are presented to the cluster as MDisks. To determine whether the selected MDisk is a flash drive, click the link on the MDisk name to display the Viewing MDisk Details pane. The Viewing MDisk Details pane displays values for the Node ID, Node Name, and Node Location attributes.

Expansion Enclosures
There are two types of available Expansion Enclosures, Storwize V7000 large form factor (LFF) Expansion Enclosure Model 12F and small form factor (SFF) 24F.
Storwize V7000 Gen2 LFF 12F includes the following components:

• Two expansion canisters 12 Gb SAS ports for control enclosure and Expansion Enclosure attachment
• Twelve slots for 3.5-inch SAS drives
• 2U, 19-inch rack mount enclosure with ac power supplies

Storwize V7000 SFF Expansion Enclosure Model 24F includes the following components:

• Two expansion canisters
• 12 Gb SAS ports for control enclosure and expansion enclosure attachment
• Twenty-four slots for 2.5-inch SAS drives
• 2U, 19-inch rack mount enclosure with AC power supplies

The Expansion Enclosure is a 2U enclosure, containing the following components:

• 24 2.5 in. drives (HDDs or SSDs).
• 2 Storage Bridge Bay (SBB)-compliant enclosure services manager (ESM) canisters.
• 2 fan assemblies. These mount between the drive midplane and the Node Canisters. Each fan module is removable when the Node Canister is removed.
• 2 Power supplies.
• RS232 port on the back panel (3.5 mm stereo jack). This is used for configuration during manufacturing.

Restriction: The Storwize V7000 Gen2 expansion enclosures can only be used with a Storwize V7000 Gen2 control enclosure. The Storwize V7000 Gen1 expansion enclosures cannot be used with a Storwize V7000 Gen2 control enclosure.

Control enclosure midplane

The midplane is a printed-circuit board, which provides power and signal connections. It contains the following blind-mating connectors:

• One SFF 8680 connector for each drive slot:
  • – Each slot has two 1x SAS 3.0 links, one to each canister.
  • – Each drive has independent power control, as supported in the SAS-3 standard (using pin P3). Midplane power field-effect transistors (FETs) per drive are not required.
• Power control is used to recover a hung drive by power cycling.
• Two canister connectors
• Connectors for each power supply

Battery backup
To retain space requirements for the IBM Storwize V7000 Gen2 and its associated components within two rack units, and to simplify the cabling and rack layout, the Storwize V7000 Gen2 node integrates the uninterruptible power supplies inside the node chassis. The control enclosure provides battery backup to support a non-volatile write cache and protect persistent metadata.

In the event of a power failure, each node performs an independent FHD from memory to the on-board SSD under software control. This method is compatible with the existing Storwize code, and it retains the data indefinitely. The memory is not required to be persistent across a
node reset.

The drives do not require battery backup, so the control enclosure removes power from the drive slots immediately before the EPOW expires (approximately within 5 ms after an ac failure). This function is implemented in hardware. Battery packs are not required in
Expansion Enclosures.

The battery pack powers the processor and memory for a few minutes while the Storwize code copies the memory contents to the onboard SSD. The FHD code runs on a single processor core to minimize power requirements. The I/O chips and HIC slot are powered down to save energy. The fans run with the node components remaining within thermal limits.

Each node switches from ac power to battery and back again without interruption. The battery supports a 5-second ride-thru delay with all node electronics active before the dump starts, in case power comes back quickly. When started, the dump always runs to completion. To allow the system to be brought online immediately after a longer power outage, the total energy stored in each battery pack supports two consecutive cycles without an intervening recharge. Each cycle incorporates a ride-thru delay plus a 16 GB FHD.

The onboard SSD has a high write bandwidth to allow the dump to complete quickly. For example, if the SSD has a sustained write bandwidth of 100 MBps, each battery pack would need to power its node canister for just under three minutes.

The operational status of batteries and their VPD are available from the IBM Storwize V7000 Gen2 command-line interface (CLI) using the saninfo lsservicestatus command.

Restriction: The Storwize V7000 Gen2 expansion enclosures can only be used with a Storwize V7000 Gen2 control enclosure. The Storwize V7000 Gen1

Below is a video tutorial that details and explains the steps to be taken.