Data virtualisation: the basis of file area networking

By Paul Phillips, country manager UK, Brocade

Storage requirements continue to grow, making it almost impossible for companies to cope. A host of factors contribute to the enormous increase in network storage requirements. These include, among others, increased integration of WANs (wide area networks) or the creation and duplication of gigantic quantities of data. In light of these factors, it is crucial that companies find new ways of efficiently organising their data and their environment.

Problems in managing enormous amounts of data are not scenarios to be dealt with in the future: administrators already face concrete problems. For example, distributed storage capacities are not put to good use. A survey by the Gartner Group reveals that a mere 30-40 percent of storage capacities in distributed networks are actually used. Set against 80 percent usage of storage capacities in mainframe environments, that is an unsatisfactory ratio. To that must be added the higher cost of distributed storage capacities. Another Gartner study revealed that mass storage and the associated administrative costs gobble up as much as 75 percent of a company’s IT budget and that for every pound invested in storage hardware, another £3.50 is spent on administration. In view of the fact that storage requirements are rising by some 100 percent annually, companies cannot afford to solve the problem merely by adding more hardware or hiring more system administrators to manage the additional burden caused by managing distributed storage systems. In addition, maintaining high availability of distributed data throughout the entire company is incredibly complex. What is required is a well designed and central network storage architecture; something that was previously not possible in distributed mass storage systems in Windows environments.

The solution in the FAN: open virtualisation
In a perfect world, a business could simply stem – or at least control – the flow of critical data, but that is a rather utopian view. More realistic is to accept that the solution must lie in lowering the cost of storage systems. Open virtualisation is the best storage management technology to do this because it lets administrators order data logically, in other words, independent of storage location. ‘Open’ here means that the virtualisation solution relies on existing standardised platforms (e.g. Microsoft Distributed File System, Domain Name System) and can manage existing and future resources regardless of the system manufacturer or technology. This means it can simply make the data available to users and applications, while at the same time separating the logical and physical data components.

So what does virtualisation mean for files? Its first aim is to shield users from the complexity of the physical storage architecture and, secondly, to enable administrators to manage this physical level without hindering user access to the data. As far as the storage itself is concerned, virtualisation is the centralisation of distributed storage systems that can then be considered and managed as individual units. Virtualisation involves grouping files from several heterogeneous storage system types (DAS, DAS+SAN, NAS) and providers. Data virtualisation gives administrators more flexibility in handling data and as well as allowing for integrated management of logical and physical data components.

Creation and management of the logical level
The logical level is often referred to as a ‘name space’. The ‘name space’ is what users see: a summarised display of the storage systems in the network presented as a single file system. It does not take into account the storage location or the number of physical devices making up the system. There may be hundreds of name spaces in a company that can all be grouped into one company-wide name space. Creating and managing name spaces is just as important as the physical administration of storage resources.

In a purely Windows environment, creating and managing name spaces is a difficult process and is prone to errors. Most organisations prefer not to take advantage of the name space architecture simply because most system administrators believe that the problems in setting up and managing a logical name space cancel out its benefits.

Data virtualisation makes it easy to create and maintain a name space of any size. Suitable solutions offer convenient functions for creating, using, managing and, if needed, reconfiguring the name space according to the user’s and organisation’s requirements.

The name space shows users the released data based strictly on function and department, regardless of its physical location. In this way, storage architectures and the respective location structures are hidden from the user. Actual changes to the physical structure of folders (migrations etc.) do not lead to errors. Thanks to the logical level, folders remain in place for users and applications, and access is still assured. This separation of the display of data from their physical locations is the first step towards effectively managing files. Once the data is referenced in the logical name space, direct access is provided between the client and the device holding the data. The direct data path avoids the additional effort that would be created by placing the logical name space between user and physical data. Because the data is referenced just once, the performance of the file system does not suffer. Bottlenecks simply no longer occur when an out-of-band data virtualisation solution is employed.

Distributed File System (DFS): the service for virtual storage solutions
DFS is Microsoft’s platform upon which other manufacturers can base their storage solutions. Because it is a component of the Microsoft Server, providers can easily develop storage virtualisation solutions using it as the basis because since Windows 95, Windows clients can be integrated in a DFS construct. The DFS service enables providers to develop, construct and manage DFS structures; and develop solutions for central storage problems, such as disaster recovery, data migration, server consolidation, user storage management, the addition or configuration of storage capacities, or for optimising available storage capacities.

In fact, DFS, data virtualisation and corresponding file management applications can solve a range of the most urgent storage management problems in large organisations. This includes tasks such as consolidating and standardising management of heterogeneous NAS devices, ongoing management of user data, data migrations, server and storage system consolidation, and – as far as possible – guaranteeing uninterrupted data availability in the event of breakdowns.

Data virtualisation optimises the administration and user-friendliness of Microsoft DFS and enables the central administration of distributed roots and logical visualisation of a complete distributed file system. There are also powerful graphics tools for configuring DFS servers and fully developed management tools for monitoring, troubleshooting and administering distributed DFS roots. Sophisticated reporting functions and web-based administration facilitate management and monitoring of DFS networks.

Many storage problems can be solved with structured virtualisation (e.g. DFS) and the management tools in the file area network that build on it. File storage virtualisation gives administrators more room for manoeuvre since it lets them independently scale and manage logical and physical storage levels. In this context, file area networking offers a whole new level of quality in reliably solving storage management problems in companies.

Brocade Communications Systems is exhibiting at Storage Expo 2007, the UK event dedicated to data storage. Now in its 7th year, the show features a comprehensive free education programme, and over 100 exhibitors at Olympia, London from 17-18th October 2007 www.storage-expo.com

•Date: 31st May 2007• Region: UK/World •Type: Article •Topic: IT continuity
Rate this article or make a comment - click here