When purchasing a virtual tape library (VTL), it's important to first develop an understanding of the features available. With this knowledge, it is then possible to create a list of features you need. Matching vendors' features to your requirements allows you to create a list of products for consideration. Features should be classified as "Required," "Nice to have" or "Not required." Those features that are not a requirement should not impact the decision process, but certainly cost is also a consideration.
By submitting your personal information, you agree that TechTarget and its partners may contact you regarding relevant content, products and special offers.
Mainframe vs. open-systems VTLs
The first criteria for deciding on a VTL system is whether your environment requires a mainframe or open-systems VTL. The uses for VTLs differ significantly in mainframe and open-systems environments and, as a result, their designs are different. Mainframe VTLs are designed to utilize disk storage to speed access to data residing on tapes. The data is actively moved between the disks and physical tape while data processing occurs. By contrast, open-systems environments typically don't process data on tapes; instead, they utilize tape for backup and archive.
Open-systems VTLs emerged a few years after mainframe VTLs. There are currently three open-systems VTL providers who OEM their software to other vendors -- FalconStor Software, Quantum Corp., and Sepaton Inc. These three providers also offer packaged appliances directly and indirectly through VARs and resellers.
There are additional open-systems vendors who deliver VTL products only as an appliance through their direct and indirect channels. These include Data Domain, Diligent Technologies Corp. (recently acquired by IBM), and NetApp. IBM and Sun both develop and deliver mainframe VTLs directly and through their channels. Diligent also has a mainframe VTL solution, which is available via Hitachi Data Systems (HDS).
The 12 major vendors delivering VTL products are listed below in Table 1: Major VTL vendors, along with the provider of the VTL software and any OEM or reseller channels for that vendor.
Table 1: Major VTL vendors
|Vendor/Product||VTL software provider||Mainframe or open systems||OEM/Resellers|
|Copan Systems||FalconStor||Open systems||None|
|Data Domain||Data Domain||Open systems||None|
|Diligent Technologies||Diligent||Open systems||HDS and Sun|
|Hewlett-Packard (HP) Co.||Sepaton||Open systems||None|
|EMC Corp. 3D||Quantum||Open systems||None|
|EMC DL4000/DL6000||FalconStor||Open systems||None|
|FalconStor Software||FalconStor||Open systems||Copan, EMC, IBM, Sun|
|Hitachi Data Systems||Diligent (acquired by IBM)||Mainframe and open systems||None|
|IBM Corp. TS7500||FalconStor||Open systems||None|
|Quantum Corp.||Quantum||Open systems||EMC|
|Sepaton Inc.||Sepaton||Open systems||HP|
|Sun Microsystems Inc. VTL||FalconStor||Open systems||None|
|Sun VSM||Sun VSM||Mainframe||None|
There are a variety of features available that are important considerations in deciding on a VTL system. Provided below in Table 2: Major VTL features and options is a list of the major features, along with the decision points and other considerations. The decision points listed in Table 2 are the questions that should be answered to determine what features are needed to meet your requirements.
Table 2: Major VTL features and options
|Feature||Decision point||Optional considerations|
|Host type||Mainframe or open systems?||N/A|
|Deployment||Integrated appliance or VTL software only?||VAR/integrator vs. internal integration|
|Disk capacity||How much physical capacity is required?||Does VTL meet needs and impact of RAID and deduplication?|
|RAID levels||Does vendor support multiple RAID levels?||RAID 6 providers higher availability with limited overhead vs. RAID 10 or RAID 50|
|Performance||Will the VTL support the speeds required for read and write?||Data dedupe, compression and other factors can affect performance|
|Number of VTL engines||How many VTL engines needed for connectivity, performance and HA?||Does the vendor support failover from any node in configs larger than two VTL engines?|
|Host connectivity||Fibre Channel, FICON, iSCSI, other?||Are multiple connection types supported?|
|External tape connectivity||How many external tape connections are required?||How many connections are needed in the future; does VTL support this?|
|Replication||Will VTL-to-VTL replication be used?||If physical tape is eliminated, replication is required for DR|
|Data deduplication||Do we require data deduplication?||Is inline or post-process data dedupe required?|
|Tape drive and library emulations||Does the VTL support the physical drives and libraries used?||N/A|
|Automated virtual to physical tape cloning||Is automated movement to physical tape desired?||Does clone operation require host or storage access?|
|Hosted backup apps||Is a hosted backup app required?||Does VTL support backup app currently in use?|
|Hardware compression||Does the VTL support HW-assisted tape compression?||Does compression match physical tape's compression?|
With the features and questions listed in Table 2, it's possible to create a comparison worksheet that would list your specific requirements for each feature, along with specifications for several vendors' VTL systems.
Key VTL considerations
Scalability is one important consideration when choosing a VTL, and it may be measured in terms of capacity, performance, number of emulation engines, number of ports and related items. The storage capacity can have a wide range, as can the number of ports and host connections. Capacity can also be greatly affected by data deduplication, which may improve the effective or virtual capacity of the system by several orders of magnitude.
The number of emulation engines or data movers can range from a single engine up to eight, and in some cases as many as 16 engines. However, larger multiengine systems may be unable to provide active/active failover among all elements, or allow management of the group as a single entity.
Data deduplication is a newer VTL feature, and there are two primary methods of providing deduplication, each with different benefits. Deduplication may either be performed inline, as the data is streamed, or as a secondary process after the data is stored. Inline offers space efficiency by not requiring additional space for un-duplicated data. Post-process deduplication requires more space, but provides higher throughput rates. Current offerings typically use only one type of deduplication, forcing users to choose which type of deduplication they will deploy when purchasing a VTL. If your backup scenario is looking to reduce backup time, then post-process deduplication may be more appropriate. If your deployment calls for moving data offsite for disaster recovery quickly via VTL replication or space savings is most important, then inline deduplication may be the best choice.
Compression is similar in concept to deduplication, though it's a much older technology. Nearly every vendor offers software compression, with many now offering hardware accelerators as well. Typically, software compression will slow down the overall data backup process. Also, the type of compression varies by tape drive type. If the VTL-emulated tape drive compression is not the same as the physical tape, then exporting a virtual tape to a physical tape may take several tapes, waste space on cartridges or the operation may fail. Thus, matching the virtual to physical compression algorithms is an important consideration.
Hosted backup application support is another option, which embeds the backup server within the VTL appliance. This provides benefits in not requiring external I/O activity or a backup host to perform a virtual to physical tape copy operation. With a hosted application, no external SAN, storage or backup host activity is required, eliminating potential contention with primary applications.
Tape caching and external tape support are related to how physical tape media may be integrated with a VTL product. External tape support allows physical libraries with tape drives to be directly connected to the VTL, so that copy operations and export operations to physical tape may occur for offsite or long-term storage. Tape caching is a concept that originated in mainframe VTL systems, whereby data flows first to the VTL, which is used as a cache, and then to an attached physical tape. With this option, there is no requirement to explicitly copy or move data to back-end tape. Instead, the data flows to tapes automatically as the VTL disk cache becomes full.
VTL replication may be an important feature, depending upon how the VTL will be utilized. For customers looking to replace tape, and maintain disaster recovery, VTL replication is essential. Other environments who are using VTL to speed up their backup and recovery processes, while still relying on physical tape for offsite disaster recovery may not require VTL replication.
A custom-designed VTL solution may provide a better match to the unique requirements of a particular data center's VTL needs than a preconfigured appliance. The software choices available allow users to deploy a system with the same VTL features as the major vendors, with more flexibility. System integrators, VARs and others may offer customized, supported VTL solutions for unique requirements. There are a variety of choices for VTL appliances as well as resellers and VARs offering VTLs; all are based on the same set of vendors VTL technology.
With all the options and considerations explained, it is possible to construct a comparison worksheet that lists your specific requirements along with several vendors' offerings. This process will help you to evaluate the key differences in VTL products and arrive at a solution that best meets your unique needs.
About the author: Russ Fellows is a Senior Analyst with Evaluator Group. He is responsible for leading research and analysis of product and market trends for NAS, virtual tape libraries and storage security. He is also the primary analyst for coverage of selected open-systems arrays and virtualization products.
Dig Deeper on Disk-based backup