Continue Reading This Article
Enjoy this article as well as all of our content, including E-Guides, news, tips and more.
Tape-based encryption also provides a measure of standardization and interoperability in the form of LTO-4, so organizations can upgrade their LTO-3 tape systems to LTO-4 while maintaining backward compatibility with existing LTO-3 tape cartridges.
LTO-4 tape technology offers an appealing option for data encryption, but there are serious issues to consider. Encryption can adversely affect compression, leading to more tape use and larger backup windows, resulting in costlier backups that are more cumbersome to produce. Tapes are also sensitive to key management, and are rendered unreadable if keys are changed or corrupted. Finally, storage professionals must continue to address the issue of tape reliabaility through proper storage and handling considerations.
Now that we've reviewed the essential issues in any encryption approach, we can focus on specific considerations for LTO-4 tape drives. After that, you'll find a series of specifications to help make on-the-spot comparisons of products from vendors such as Hewlett-Packard Co., IBM, Quantum Corp., Sun Microsystems Inc. and Tandberg Data.
Consider the impact of encryption on media costs and storage expenses. Virtually all tape systems run in compressed mode for maximum capacity and throughput. A standard LTO-4 tape can double its native (uncompressed) capacity of 800 GB and 120 MBps throughput to 1.6 TB of compressed capacity at throughput up to 240 MBps. However, encryption effectively scrambles the data, removing the redundancy that compression algorithms rely on.
If you are evaluating a tape drive with tape-based encryption, ensure that compression will be performed prior to encryption. Otherwise tape data will not compress and each backup cycle will use more tape (and take longer), using more costly media. Drives such as HP's StorageWorks LTO-4 Ultrium 1840 and Sun's T10000 tape drives implement hardware compression before encryption.
Consider software support for tape-based encryption. Tape-based encryption is not an automatic feature. Applications such as backup software must be able to support the drive's encryption feature, but broad support from independent software vendors (ISVs) may not appear until early 2008. In the near term, storage administrators may find themselves in a quagmire: Either forego tape encryption until their backup software products are updated to support the new drives, or forego their existing backup software in favor of the backup tools bundled with the drive.
Consider drive features that improve performance. Tape-based encryption should have little (if any) noticeable impact on drive or backup system performance. But the drive's features should help mitigate any performance mismatch between the drive and backup server. If the backup server is too slow and cannot stream data to the tape drive continuously, the tape drive must stop, reposition and start again. Such "shoeshining" is a common behavior, but it can be detrimental to the tape drive and media. Today's advanced tape drives offer large buffers and can adjust their transfer rate and internal speed to match the server. HP calls this feature data rate matching (DRM), allowing its tape drives to dynamically adjust from 40 MBps to 120 MBps.
Weigh the importance of tape drive standardization. Not all encryption-capable tape drives are fully compatible with the capacity or compression levels called out in the LTO-4 standard. For example, IBM's Tape Drive Express TS2340 specifically calls out the LTO Ultrium 4 standard at 800 GB capacity and 120 MBps throughput (uncompressed) and 1.6 TB capacity and 240 MBps throughput (compressed). However, Sun's T10000 drive touts 500 GB capacity and 120 MBps throughput (uncompressed) and promises up to 3-to-1 compression. Although tape cartridges that are written on an LTO-4 drive should be readable on any LTO-4 drive, those tapes may not be readable on non-standard drives (and vice versa). Encryption should also be standard. For example, a tape encrypted in an LTO-4 drive should be readable in another LTO-4 drive as long as both drives support LTO-4 and a suitable key is supplied.
Consider how the key is stored and used. A key is needed to encrypt the data, but a key is also needed to recover the encrypted data. This usually involves storing the key where it is accessible to the tape. Some drives store the encryption key directly on the tape, and the key itself is encrypted with public key cryptography (asymmetric keys). This makes it possible to share encrypted tapes between trusted partners or multiple data centers without having to transmit keys. In other cases, the key is stored in a dedicated "key management" appliance, and a key identifier is placed on the tape. When that tape is read, the tape requests the corresponding key from the appliance. Consider how the key management system adds security to the organization, but also evaluate the level of complexity, cost and the effect that any future hardware changes or disasters might have on the key management process.
Consider support for the operating system. New devices typically require a corresponding device driver, so the addition of an LTO-4 encryption-capable drive should also include a device driver that supports the operating system on your backup server. When considering a new drive, check to see if drivers are also available for current/future operating systems. This will help to ensure that the drive will continue to operate properly as operating systems are upgraded into the future.
Consider support for WORM media. Data that is recorded for long-term archival and compliance/litigation purposes may require immutability -- meaning that the data cannot be deleted or altered once it's written. Many current-generation tape drives support write-once read-many (WORM) media. WORM media typically uses memory on the tape cartridge and a unique encoding scheme to prevent data tampering.
The tape-based encryption product specifications page in this chapter covers the following products:
This was first published in September 2007