File Virtualization Evolving: Q&A with Klavs Landberg

By Carryl Roy
published: Tuesday, January 19 2010

KL: You can use server virtualization to put multiple Windows operating systems on a single computer.  This allows you to consolidate application servers and reduce physical device sprawl.  But server virtualization doesn’t help much with management complexity.  

With file virtualization, you can reduce both physical devices and operating systems without changing the way that files are presented to end-users.  You can consolidate a large number of small file servers to a single file server.  You manage just one logical system – one piece of hardware, and one operating system.  The file virtualization takes care of client referrals to the actual location of their data, and clients continue to see their files in their original location.  

File virtualization does not stop there.  Now you can move data freely – you can tier it, replicate it, or archive it.  Bottom line: Server virtualization is for consolidating applications.  File virtualization enables and provides a complete storage management solution.  

KL: Sort of.  I have always thought of AutoVirt as the integrated combination of file virtualization and policy-based automation.  File virtualization is a way to automate storage management and make it non-disruptive.   File virtualization creates a separate data model that is designed to fulfill the needs of end-users, and at the same time, to provide IT complete independence to manage data and storage whenever and however they need.  

So is AutoVirt a file virtualization or a data management solution?  Both.  The focus depends on who you’re talking with and what their frame of reference is.

KL: Lots of people still don’t realize what file virtualization will do for their environment.  File virtualization is similar to DNS in a very direct way.  The common purpose is to help users navigate their networks and find the resources and data that they need to do their jobs.  File virtualization does for file access exactly what DNS does for browsing.   

File virtualization is unique in that it separates access paths from the physical location of networked data.  In storage environments where data must be moved – and this is the case in most environments today – a literal interpretation of file server and share names is not much better than people memorizing IP addresses.  For this reason, file virtualization and global namespaces are as inevitable as gravity.  

People also don’t realize that file virtualization can be installed quickly and seamlessly – without any changes to the way that users see their networked files, and without manual work for IT.  

KL: Successful file virtualization is provided by an out-of-band solution that installs quickly and easily.  It must provide perpetual support of pre-virtualization UNC names, scalable performance, integrated data management, and compatibility with existing products and processes in the IT infrastructure.  

KL: No IT organization can afford to break already working services.  When you install a new product, you shouldn’t have to recreate your entire client access naming system, which is what some file virtualization products require you to do.  Consider breaking user access to all files on the network at once – it’s a prohibitive process.  You need to be able to support all existing UNC names found in shortcuts and embedded links.  Seamless insertion is like the ante in poker: pay up or leave the game.

KL: Generally speaking, out-of-band solutions scale much better than in-band solutions. I define scalability as the ability to add functionality without new limitations to size, performance, or locality.  For file virtualization specifically, size applies to how big the namespace can be and what tools are used to host it; performance applies to additional latency and incremental network traffic; and locality applies to the ability of the namespace to be present in all required locations.

KL: The true purpose of file virtualization is to allow data management at the same time and independently of user access to the same data.  With file virtualization, you have the freedom to manage data much more rigorously – which is why a file virtualization solution should be closely integrated with a data management platform. 

KL: CIOs must plan to handle data that is growing 10 times in volume every 5 years.  Fortunately, disk storage is getting cheaper – so it’s possible to add capacity as it is needed.  File virtualization allows you to quickly and seamlessly handle additions of new data and new storage.  File virtualization is also a great way to deal with changing legal compliance and corporate governance requirements, because it enables fluid tiering strategies.    

For the IT administrator, the most important thing is to keep the IT infrastructure up and running every single day.  The great thing about file virtualization is that it can be inserted seamlessly, with very little impact on day-to-day operations.  Once installed, you can use it for routine tasks like replication or small migrations, or you can use it only for large-scale, transformative projects.  A good file virtualization product will coexist peacefully with your existing infrastructure, so you can use it to whatever extent is convenient and necessary.

Related Links:

AutoVirt