File virtualization can address storage capacity issues and benefit storage infrastructures containing data that needs to be accessed by multiple applications. However, complications with file virtualization can arise
You must have Adobe Flash Player 7 or above to view this content.See http://www.adobe.com/products/flashplayer to download now.
Download for later:
How file virtualization can benefit your storage infrastructure
• Internet Explorer: Right Click > Save Target As
• Firefox: Right Click > Save Link As
Table of contents:
What is file virtualization?
Different file virtualization technologies and how they vary
How file virtualization affects replication or migration of files
Benefits and drawbacks of file virtualization
Products that currently enable file virtualization
Boles: File virtualization is really about separating file data from restrictions, and how the data is accessed and manipulated. You can get a lot of functionality out of accessing files via CIFS or NFS, like you probably do in your network-attached storage (NAS) environment, whether on a traditional file server or on a NAS appliance. But at the same time in this file environment, you get roped into a lot of restrictions. Most importantly, your clients, desktops or servers can get tied into a single access path, and if you try to move files or directories, you can break things pretty badly.
File virtualization can mask the way you access your file data and, in turn, let you move and manipulate your storage files without impacting your clients. So if your storage infrastructure has really grown and you have different data being accessed by different business systems, this can be a tremendous benefit. It can also ease things when it comes to migrating storage systems. This is where we come to some complexities with the assortment of file virtualization technologies available on the market today.
Boles: While the capabilities I've talked about so far are the core focus of most vendors, there are lots of other capabilities that get lumped in the file virtualization space. The capabilities I've talked about so far are what the Taneja Group has commonly referred to as network file management. With this, you will typically insert a device in-band or integrate an out-of-band device with your NAS storage. Those devices will move files and either actively redirect clients or leave a stub file or pointer behind that the client follows to a new file location. Network file management is what users generally think about when referring to file virtualization.
Behind this concept, users should recognize that some solutions do file-based virtualization and some do directory-based file virtualization. The choices vary by how much horsepower they require and how granularly they let you tier and manipulate data. So with file-based, you can virtualize down at the file level, but you may need more horsepower depending upon whether it is an in-band or out-of-band solution. You'll also find different selections of solutions based on whether they are file- or directory-based, and in-band or out-of-band. In some cases, you may have a choice of three or four solutions that are in-band and file based, but there are no solutions that are out-of-band and directory based.
Some technologies may be more appropriate for doing periodic migrations while others may be something that you leave on, so you can always tier and manipulate your data. These technologies permanently virtualize your environment, whereas some other technologies may only cut in and help you migrate data.
We should also talk about global namespace here, as this technology is often lumped in with file virtualization. Global namespaces can aggregate all of your different files and directories into one point of access, so you don't have to independently manage file points, NAS systems and client configurations.
Boles: Depending on the solution, file virtualization can interact with your file server in a number of ways. The solution might be integrated with NAS APIs and act on the data behind the scenes to redirect a client to a moved file. Depending on how you replicate and migrate your data, it may complicate your environment. If it is done strictly by a file system and you have tools that apply to a file system, you may not migrate and replicate that set of data in the same way. So you need to think about that when you're determining how you're going to architect one of these solutions into your environment.
Similarly, if you're doing file virtualization in-band, you may introduce some latency into your environment. You may also need to think about how you're handling failover and migration of new solutions. In addition, this will have the same impact on replication and migration.
While you can apply these solutions to do a migration, if you're using replication and migration technologies outside of these solutions, there may be serious implications. So you need to think holistically about the file SAN environment you're going to architect with these tools.
Boles: File virtualization has been designed to tie storage silos together and make management and storage optimization easier by putting the right data on the right tier. Because storage pros tend to provision extra file storage capacity, it eases infrastructure requirements to allow for lots of extra capacity and overhead so that we don't get forced into a migration event.
If you do run a file system up to 95% of its capacity, you may be at a point where you have to do an emergency migration event. With file virtualization, you can move some of that data and run your system closer to that maximum capacity limit to avoid an emergency migration. So you should to go into a file virtualization solution with an eye on your overall storage architecture.
As I mentioned, there may be implications down at the replication and migration layers of your infrastructure, but there also may be some tie-in to your file virtualization solution making it difficult to back out of. If you're doing a whole lot of granular file tiering, you may find that it's challenging to un-tier that data and take that solution out of the infrastructure. You need to think about these things in the long term and how you're going to migrate away from that environment should something happen to the vendor or solution.
Boles: We have the usual powerhouses, like EMC Corp.'s Rainfinity and F5 Network Inc.'s Acopia, that really helped define this market. Both of those solutions have the full range of capabilities that you would expect from any file virtualization suite, including global namespace capabilities, file-level tiering and directory-level tiering, and they give you a choice of in-band and out-of-band approaches.
Trickling down to the small- and midsized business (SMB) market, you have players out there like AutoVirt [Inc.] that came to market with a solution targeted more at solving the SMB's file migration challenges. There is also a company out of Austria by the name of SmApper Technologies.
When you start thinking about what you can do with file virtualization outside of just a tiering migration capability, the possibilities become really interesting. File virtualization gives you a foundation on top of which to build policy engines surrounding flat data, and the ability to cache in a low-balancing failover application integration. It's actually those possibilities for which the term file area networking applies.
This was first published in September 2009