What to know for your virtualized storage selection process

Types of storage virtualization

There are two main storage virtualization types: file-based and block-based.

File-based storage virtualization specifically works with network-attached storage (NAS) systems. It breaks the connection on a NAS array between data access and a memory's physical location. File-based virtualization simplifies background file migration and improves performance.

More common is block-based -- or block access -- storage virtualization, which uses Fibre Channel or an iSCSI storage area network. These systems abstract the logical storage from a device's physical memory blocks and pools all available memory blocks into a virtual, shared resource. Block-based virtualization has less of a read/write overhead, which automatically improves overall performance.

Storage virtualization methods

There are five main methods of storage virtualization: host-based, array-based, OS-level, file system and Fibre Channel. These all abstract the storage differently, which can determine the best option for your organization.

Host-based storage virtualization. You run a software layer on the host layer to pool storage resources and present them to a guest OS. The original host runs a privileged task or process to intercept any I/O requests, run I/O mapping and provide metadata lookup, then allows virtual storage access. This method does not require a specific type of host or device.

Array-based storage virtualization. The storage array is the main storage controller and runs the virtualization software. This method lets you get the power of flash storage without the purchase of an all-flash array; the system reads pooled storage as different tier types. It can write all data to the high-speed tier and then automatically move it to the standard tier.

OS-level storage virtualization. This method has similar capabilities of array-based virtualization but has them at the software level. It runs multiple user spaces, such as containers or virtual private servers, on the OS. Certain OSes, such as Windows Server Storage Spaces, have built-in virtualization features.

File system storage virtualization. This provides a consolidated look at file data. You might think that your data sits on a single file server, when it is actually distributed across multiple file servers.

Fibre Channel virtualization. This is an overarching term for virtualized storage types that use Fibre Channel. Also known as network-based storage virtualization, this approach uses Fibre Channel switches to redirect I/O requests to physical storage without a server consuming any storage.

Virtualized storage vs. software-defined storage

You may think that the terms storage virtualization and software-defined storage are interchangeable, but this is not the case. The difference between these two technologies is how they provide and initiate storage access.

Storage virtualization combines storage pools into one or more logical containers to consolidate resources. This approach allows you to implement storage-based management policies, have more control of storage at the VM level, improve performance tracking and potentially automate compliance.

Software-defined storage shifts the hardware layer that provides deduplication and snapshots into the software layer. This model means there is more control for application-centric management in terms of storage redundancy, capacity and performance. It also lets you refresh older storage infrastructure without storage frame replacement.

Considerations for virtualized storage selection

Once you decide to explore storage virtualization options, there are some factors that you must be aware of to effectively vet software.

Understand the business need for virtualized storage. This way, you avoid paying for more performance than necessary. Potential reasons for storage virtualization include faster VM spin up times, cost reduction, outage prevention or reduced capacity for data deduplication.

The move to storage virtualization also adds complexity, so ensure your IT team can take on the additional moving parts and understands the technology before purchase and implementation.

With the technology, assess your infrastructure to avoid incompatible virtualization layers. Lack of interoperability between arrays or virtualization layers increases complexity and can potentially be a waste of money.

Evaluate your performance requirements to ensure that the method and hardware of the virtualized storage setup matches bandwidth requirements; iSCSI is likely fine for an SMB, but an enterprise should look at Fibre Channel SAN.

Determine what reporting workflows you require for efficient infrastructure management -- and if the virtual storage installation will interrupt those reporting structures. Tools that work well for nonvirtual storage hardware reporting and metrics might not fit in virtualized setups or have the same capabilities.

Be sure to think long term. If you implement virtualized storage, know if the initial upfront cost is worth it in the long run and your organization can also handle the long-term cost of virtualized storage.