Let's start by challenging preconceived notions of what storage virtualisation might be, and accept this broad definition instead: "a storage system that employs virtualisation is one that abstracts the relationship between the volume presented to the server and a group of specific disks". Sure, it may irk the old-school database administrator that you can no longer point to the 10 disks that hold the ERP system; however there are considerable benefits for moving away from this arrangement.
By the way, if the definition of storage virtualisation sounds suspiciously similar to that of a consolidated storage system, or SAN, it is because the lines between the two continue to be blurred.
Today, functions that enhance the value of traditional offerings must be developed simply to maintain market share. Arguably the only place in a modern enterprise for a featureless storage system with solid performance characteristics is beneath a storage virtualisation layer or as a backup-to-disk repository. So, rather than restricting the consideration to only certain types of storage virtualisation, a wider definition allows us to fully consider the benefits you ought to expect to gain if you deploy storage virtualisation for your clients. It is the benefits, after all, that are the true motivations for deploying storage virtualisation, rather than the accruing of a magpie's collection of bright and shiny new charms, is it not? A storage virtualisation platform is a prerequisite for even modest server virtualisation projects.
As you read the benefits of storage virtualisation below you will see how they mirror those of server virtualisation, maintaining the principles of establishing a dynamic infrastructure. The benefits are all aimed squarely at reducing costs while improving service levels – a good premise for any IT system. Naturally, not all benefits are available from all storage virtualisation systems, but here are some of the benefits you can expect to see:
Improved utilisation and performance: This pools together many disks, arrays, or even several disk systems, then thinly provisions logical volumes to eliminate the classic problems of high use and hot spots in some areas whilst having low use in others.
Increased availability: This eliminates downtime associated with data movement through online data migrations. Use those same online data migration methods to eliminate maintenance-window outages and simplify end-of-life and end-of-lease migrations. Better manageability: Provide a single point of administration and management for capacity and performance in the system, in order to reduce complexity and costs.
Ideally this is built into the system without requiring additional components.
Freedom of choice: This draws on commodity components as the building blocks of a storage virtualisation system to not only reduce cost, but also to consolidate high business-value functions like copy services into the virtualisation layer. This has the added benefit of enabling those functions across multi-vendor environments, where this was previously not possible. Other benefits include scalability through scale-out design, higher degrees of risk mitigation through server virtualisation integrated remote copy services, and increased redundancy through multi-node deployments.
The storage virtualisation approaches available today can be categorised as follows:
Multi-tier storage systems (eg: SSD, SAS, SATA) using traditional RAID arrays and functions to stripe across them; thin provisioning; copy services between same or similar devices; and in some cases automatic migration of volumes or segments between tiers based on usage policy. These systems are characterised by a high degree of control and performance within a system.
Single-tier systems that dispense with traditional RAID arrays and tiering by using massive grids of disks, cache and bandwidth simultaneously from commodity components.
These systems are characterised by ease of configuration and use, scalability and redundancy. Storage virtualisation layers built into network-based appliances or within storage controllers (both block-based and NAS gateways) that pool the capacity of multiple systems and provide online migration and copy services in heterogeneous environments. These systems are characterised by tremendous performance, flexibility and scalability. Other virtualisation approaches using custom configurable Windows-based virtualisation appliances; easily deployed virtual server appliances; managed reliability systems that automatically re-initialise failed HDDs; and multi-node network RAID protection systems.
These systems are characterised by providing a subset of functionality at a proportion of the cost of more comprehensive systems. The storage virtualisation approach selected should be the one that matches the benefits the organisation needs and desires. This should be grander than a comparison of features and performance and may be more about aligning with the culture of an organisation, its desired level of control over its systems, its previous experiences both pleasant and painful, purchasing cycles and authority, and maturity of IT in other areas of the organisation like server virtualisation and growth planning.