Guidelines for evaluating virtual tape libraries

VTLs can form the foundation of a multi-layered data-protection strategy.

By David Hill

Virtual tape libraries (VTLs) deliver faster and more-reliable backups and restores, but that’s not necessarily a reason to buy them. Although restoring from a VTL is faster than with a physical tape library, VTLs may not qualify for high-availability applications with stringent recovery time objectives (RTOs). Other alternatives, such as remote mirroring and continuous data protection (CDP), may be better suited for these requirements.

Nevertheless, VTLs can play an essential role in breathing new life into traditional backup/restore environments. Standard backup is not the first line of defense for data protection. For physical data protection (e.g., against disk failures), the first line of defense is RAID and/or remote mirroring to a disaster-recovery site. For logical data protection (e.g., against viruses or accidental file deletion), snapshots or newer approaches such as CDP offer a first line of defense.

However, many enterprises require several layers of data protection-even at the price of lower availability of data. Tape automation serves as the last line of defense, with media at both remote and local sites in a media rotation strategy.

Tape automation has been under increasing pressure due to ever-expanding amounts of storage and requirements for applications to be available 24x7. One clear example is the “running-out-of-night” problem, where backups cannot be completed during the time available. The addition of VTLs relieves those pressures and extends existing backup/restore processes, including tape automation.

Why rejuvenate instead of euthanize the old environment, since the use of VTLs implies that an organization plans to maintain its current backup/restore processes-including tape libraries? Although alternatives exist that may be appropriate in certain situations, IT organizations have to be cautious about taking a revolutionary approach where failure could be dangerous to critical business data.

VTLs are an evolutionary step forward, offering a platform that can be smoothly assimilated and can measurably improve data-protection environments. VTLs can also serve as a centerpiece for data protection while working with other data-protection technologies such as CDP or remote replication to a disaster-recovery site. In other words, the last lines of defense can work in conjunction with the front lines of data protection.

This is important since VTLs are typically additions to a data-protection environment and not a replacement for existing investments. But even though VTLs add to the capital budget, an ROI case can be made for the investment.

Plenty of vendors

The potential of VTLs has not escaped the notice of vendors (see table on p. 38 for a comparison of selected VTLs). Both the Goliaths and Davids of the industry are well-represented. However, not all VTLs are created equal. A key line of demarcation lies between VTL appliances and software-only solutions.

A VTL appliance is a turnkey platform that includes the software that drives the VTL functionality, the server on which the software runs, and the disk array that stores the backed up data. Some vendors loosely term the VTL software and server as a VTL appliance, which fits one definition of an appliance as a dedicated-purpose server. However, we believe IT organizations interested in buying VTL appliances are best served by VTLs that incorporate storage arrays. Perhaps the term VTL gateway (a la NAS gateway) would help differentiate the two alternatives.

On one hand, the higher level of integration that comes with a VTL appliance that includes a disk array eases the burden of initial deployment as well as on-going administration issues. On the other hand, a software-only solution (i.e., VTL gateway) allows companies to use existing server and storage investments and does not dictate a proprietary solution, thus increasing flexibility and providing investment protection.

Other considerations for VTL product selection include the following:

Is the VTL compatible with your existing operating systems, servers, backup/restore software, tape formats, and tape libraries?

Does the VTL meet the IT organization’s reliability, availability, and serviceability (RAS) standards?

How easy is it to configure, tune, and manage the VTL?

What is the data movement process between the VTL and a physical tape library, and how does this affect how restores are performed?

How is compression handled, both on the disk array and tape media? How does the system manage incremental backups? What data reduction techniques, if any, are used if electronic vaulting to a remote site is supported?

What is the file restore granularity? Can single files be restored easily (i.e., no volume restores necessary) and, if so, can the restores be done by a user (with the proper authentication)?

What value-added functionality beyond the basics does the VTL offer, such as electronic vaulting capability to remote sites or integration with other data-protection technologies, such as CDP?

Click here to enlarge image

One of the biggest issues with VTLs relates to moving backup copies from disk to a physical tape library. There are basically two approaches:

The backup server controls the movement of data from the VTL to the physical tape library. Although the physical tape library is visible to the VTL, the backup server controls the process. The advantage of this model is that if the backup that the backup server needs is not in the VTL, then the server has the correct catalog information to be able to restore from the physical tape library. The disadvantage is that the backup server has to use resources to make a copy that the VTL could have just as easily made in the background.

The VTL simply performs a copy-and-export function to the physical tape library. Technically, this is the preferred solution since the backup server does not have to get involved in the duplication function. Practically speaking, however, having two backup copies that are identical (i.e., have the same bar code) can create consistency problems for the backup server. For example, if the backup server were not given the information that a tape copy returned from off-site was the correct copy, then the backup server would assume that the correct copy was in the VTL and reject the physical copy. Workarounds exist, but workarounds require compromises.

The Virtual Tape Library Special Interest Group (VTL SIG) within the Data Protection Initiative (DPI) of the Storage Networking Industry Association (SNIA) is working on a general solution to this issue and others related to VTLs. That work constitutes a positive step forward, but should not inhibit IT organizations from evaluating and implementing VTLs today.

However, IT organizations should not rush to deploy VTLs just because the strains of trying to manage current backup/restore processes are exacting an unacceptable toll.

Rather, they should take their time to determine which solution is most likely to relieve the immediate symptoms and also serve as a foundation for building more effective data protection infrastructures.

David Hill is a principal with the Mesabi Group (www.mesabigroup.com), a consulting firm specializing in storage networking and management. A version of this article originally appeared in the Pund-IT newsletter, www.pund-it.com.

This article was originally published on May 01, 2006