Virtualisation of IT systems enables admin staff to run datacentres effectively while hiding complexity from end-users, says Hamish Macarthur
Systems that operate simply and intuitively are built on sophisticated technologies. Virtualisation is one such technology, such as when several users connect over the same communications line. Virtual memory within a server is another example, where many applications or tasks can share or use the same physical resource after this has been divided into separate virtual "containers".
So what does virtualisation offer? Virtualisation enables a system's physical resources to be shared by different tasks and so it helps achieve higher utilisation of the assets.
The reason why virtualisation is a key component of most IT infrastructures is because it enables systems to operate more effectively. Virtualisation also allows users to react to system and component failures so the rest of the system does not fail.
The continuous development of the system architecture will result in a computing grid. For this to operate, virtualised services must be able to work at many levels - sharing resources, enabling safe failover, restoring data to a trusted state and providing self-healing characteristics.
Overall, virtualisation of technology plays a significant role in making systems operate smoothly and with little degradation to end-users.
There are various storage implementations of virtualisation. Technologies such as Raid, volume management, disc array virtualisation and virtual tape are all examples of virtualised storage resources.
Raid is designed to protect data loss when a disc drive fails. Implementation of different Raid levels will determine how quickly the data stored on a disc can be reconstructed to support continuous system operation.
If Raid 0/1 is used, then there is a complete copy of all data on another set of discs ("mirroring"). This means that data loss can be handled as a background task and all systems operate off the second image of data. Developments of Raid 6 will automatically recover from two discs failing simultaneously.
When data is written to disc or tape, it is desirable that all relevant files or databases are kept in an orderly fashion, in volumes. To enable the data to be handled across disc drives and disc arrays, volume management was developed. This means that there is no longer a boundary point once a disc array is full of data. Volumes can now be placed across disc arrays and still be managed as a single entity. The application is now just viewing the data, wherever it is physically located.
In storage networks, it is desirable to use existing investment in different disc arrays or different suppliers' arrays. Despite the fact that these arrays feature their own characteristics, storage virtualisation enables volumes of data to be managed across heterogeneous arrays.
This virtualisation software can be located at the network level on an appliance or it can be a feature of the disc controller.
Virtual tape is yet another example of storage virtualisation. When backing up to tape, the time taken is dependent on the tape speeds as well as the number of drives. But when it comes to restoring data, then restore times can be extended when downloading the data from tape.
Virtual tape products use the same process as writing to tape, but instead the data is physically written to disc, making restore times faster. For long-term archival, the data is written to tape from the disc cache which is emulating tape.
The technologies listed are just some of the examples of storage virtualisation. For instance, with many network attached storage (NAS) devices in a network, it becomes prudent to rationalise the access to all these devices. This can be achieved through a virtual NAS head that then directs the transactions to the relevant devices. Server virtualisation is also important as end-users seek to gain more out of their investment.
The real gain for companies and reason for virtualisation technology is to enable more complex systems to operate reliably and continuously, masking complexities from operators and end-users.
Unquestionably virtualisation will be of increasing importance in all systems, not least when we look to the development of computing grids within the datacentre and across sites.
Hamish Macarthur is chief executive at Macarthur Stroud International