Should virtual tape libraries (VTLs) be the servant of backup software and replicate the physical tape library down to individual tape reels and labels or not? FalconStor's technology VP, John Lallier, can see a role for a VTL as a tape substitute, in which case it has more freedom as a media manager. In the tape substitution area extending a VTL's capacity is going to be hugely important. De-duplication appears to be the most promising way to do this. But where should the de-duping take place? Should be at the data ingestion point or after ingestion? Falconstor's view is informed by the need to preserve a VTL's ability to dramatically reduce backup windows.
TechWorld: FalconStor is one of the largest providers of virtual tape library (VTL) systems, both under its own brand and through several very large OEMs. As a leader, how does FalconStor define a VTL?
John Lallier: A VTL is a disk-based storage system that appears to the outside world to be a tape library with two or more tape drives. It is important that it mimic a tape library and not just a tape drive in order to provide the scale in capacity as well as concurrent access. The key is that for any backup application, this system should be indistinguishable from an actual tape library.
TW: What about those who insist that the VTL needs to do more than emulate a tape library – that it needs to de-couple the virtual tape from its physical counterpart?
JL: The key to a VTL’s success is that it is non-disruptive. It performs disk-to-disk backup without the need to replace all the existing disk-to-tape backup components. And the least disruptive way to use a VTL is as the higher-speed stage for backups that will eventually be moved to physical tape, exactly as the application wrote them.
For a large percentage of our customers, the backup application has become the de-facto media manager. All backups are performed by the backup application, and its database is the catalog for what data is on which tapes. For these customers, it is important that the data the backup application records on LTO2 tape TPE20070101 is actually on the physical LTO2 tape TPE20070101. If that physical tape is put into any tape library, the backup application should be able to retrieve the data directly, without relying on the VTL to “translate.”
TW: So there’s no need to de-couple the virtual tape from the physical tape?
JL: Not for these customers. But there is another set of customers that are using the VTL as a substitute for tape. For these customers, only a fraction of the virtual tapes will ever be transferred to physical tape. Other customers make use of the VTL’s ability to present the backup application with one type of media (9840, DLT) and then transfer the virtual tape to a physical tape of a different type (LTO, SDLT).
Finally, there are large customers that are looking for the open-system VTLs to have the same capabilities as the mainframe virtual tape systems. For these customers, the backup applications are not the media managers, and the expectation is that the VTL will organize the disk storage and tape storage automatically based on the rules laid out by the customer. FalconStor is working with our partners to enhance our products to provide this level of service.
TW: So what is the big new thing in VTLs?
JL: The biggest request we’ve seen from our customers is in the area of capacity, specifically using de-duplication to increase the number of backups that a VTL will hold. Backup data is the ideal candidate for this type of approach since the backup data tends to have a lot of redundancy over time.
TW: But many companies are working on this. How does FalconStor’s approach compare?
JL: The most important factor we looked at with de-duplication was to not impact the current advantages of the VTL. Most people are using VTLs to get more backup data processed in very limited windows of time. Anything that would adversely affect the VTL’s performance would be unacceptable.
We believe the right approach is to do the de-duplication after the backup has completed. This way, the systems you’re backing up are free to be used for their primary purpose – running your business. And since this is de-duplication of backup data, there is an advantage in recognising that fact and making use of content-aware de-duplication rather than relying on a brute force method.
Find your next job with computerworld UK jobs