DRJ Fall 2019

Conference & Exhibit

Attend The #1 BC/DR Event!

Fall Journal

Volume 32, Issue 3

Full Contents Now Available!

combining_1One of the keys to any data recovery planning and implementation is selecting the right tools to recover the data. Once the key people, data, and plan have been identified, the right combination of hardware equipment is required. In virtually every situation, it makes sense to include tape as part of your equipment selection.

This may not be consistent with conventional wisdom for data recovery, but it’s true nonetheless. The disk industry took the lead in data protection following poor tape performance in the 1990s. The old perceptions about tape have lingered, but the fact is that tape reliability has improved more than 700 percent over the last 10 years and is now quantifiably more reliable and faster than the SATA disks that are typically paired with tape. This speaks to a significant point: just as disk has a key role in data protection and disaster recovery, so does tape.

A Little History

Tape had some rough years, and this is an undeniable fact. During that time, disk started assuming an important role in data protection. In most data centers, disk has been used as a step in the back-up and recovery process so that data moves from primary disk (such as Fibre Channel disk) to secondary disk (typically SATA) then to tape – disk-to-disk-to-tape (d2d2t). This was made possible in part because SATA disk is more affordable than Fibre Channel disk and much more so than solid state disk. Because SATA precedes tape in typical data protection tiering, SATA disk is considered roughly equivalent to the tape tier and used when comparing the technologies.

Best Use of Disk in Data Recovery

SATA disks’ two strengths in terms of data backup and recovery are random access of information so that specific files can be rapidly restored and a familiar interface that allows users to access files without requiring proprietary software to read the data. Disk is hands-down the best choice in tiering data, especially using SATA for the second tier of storage. The short-term use of disk in the data protection schema is a significant advantage.

Analyst studies indicate that 95 percent of data recovery occurs in the first two weeks following data creation, making disk the ideal short-term storage medium. However, disk is much more costly than tape, so tape is typically a better choice for longer-term storage. Just as shorter-term storage plays to disk’s strengths, longer-term storage plays to tape’s strengths.

For sites that manage data carefully, deduplication and remote replication to disk have some advantages, primarily through the ability of the site to restore the most recent data. Note, however, that for large quantities of data or sites affected by power blackouts, this advantage disappears. Copying terabytes of data over a WAN is time-consuming and expensive. With tape, sites that have a great deal of data to restore can move that data more quickly, even if it requires overnight transport to a failed data center.

Meet the New Tape

In spite of a common perception of disk as extremely reliable, SATA disk is, in fact, a significantly less reliable technology than tape. Given the bit error rate of each technology, it turns out that “…you are 100 times more likely to have bad data on disk than you are on an LTO-5 tape drive and 10,000 times more likely than if the data is stored on a T1000C or TS1130 drive.” New tape library technology has advanced tape’s data integrity capabilities, offering features that verify data integrity once data has been written to tape. With these technologies, automated tape libraries offer the most reliable storage tier — more reliable than SATA disk by at least two orders of magnitude.

Tape is also faster than SATA at moving a lot of data — and obviously this is important in getting an organization running again after a disaster. When comparing throughput, drive to drive, tape drives have proven to be faster than SATA drives. One of the fastest SATA hard drives in a recent benchmark study touted a SATA drive’s throughput of 157 MB/s. This throughput is significantly slower than tape, with the newly released T10000C drive posting uncompressed data throughput of 240 MB/s, and the TS1140 posting uncompressed data throughput of 250 MB/s. Tape’s burst transfer rates are even faster, and compressed data posts throughput speeds of up to 500 MB/s. This compares well with even Fibre Channel drives which support 400 up to 800 MB/s throughput speeds.

Tape is the most cost-effective storage option for data that can tolerate access times of a few minutes. This is particularly significant when administrators consider the amount of disk that is used appropriately. (See above graphic.)

Studies show that about 30 percent of the data stored on disk is appropriate, while “about 40 percent of the data is inert ... 15 percent is allocated but unused; and 10 percent is orphaned, meaning the owners of the data are no longer with the organization; and 5 percent is inappropriate.” Given that, on average, only 30 percent of data on disk is appropriately stored for wider use, the relative cost of disks is further inflated as compared with tape.

combining_2Tape’s Drawbacks

Perception is tape’s single biggest problem, although that is not to say that tape is trouble-free. In locations where legacy tape systems are in use, problems are common. In centers using current technology, tape problems are few, but tape still requires at least minimal best practices, just as all IT equipment and processes require proper care. At well managed sites, such as the National Energy Research Scientific Computing Center (NERSC), where tape is used heavily, the success rate of data restore is 99.945 percent. This includes restoring tapes up to 12 years after data is written. If nothing else, this confirms and provides proof that tape can perform very well.

The time required to restore individual files is significant if the tapes are not in the library – a classic case where using disk simply makes sense. Tape that is in the library has a latency of only a few minutes. Latency issues are largely removed with the market’s increasing adoption of active archive, where a copy of all data is kept in the library.

Networks and back-up applications are currently optimized for disk, not tape, so it is difficult to get the full value of tape’s speed advantage. This is another case where using disk as an intermediary makes sense – d2d2t, with disk between primary storage and tape.

The proprietary interface to data stored on tape is another hindrance to data restore. It is important to note that this is beginning to change, with the release of Linear Tape File System (LTFS). With this file system interface, tape can be mounted and appear as a disk drive. Users can drag and drop files without having to go through proprietary software that translates data formats to human readable versions.

Some would say that tape is untenable due to the overhead of having to migrate data to newer generation tapes or having to have copies of old operating systems and software. However, this assertion overlooks the fact that data must be frequently migrated from disk to newer disk generations. The average life of disk is about four years, with some suggesting three years as the optimal time between disk replacements. This compares unfavorably with tape, which has a typical lifespan of up to 12 years. Tape is particularly flexible in that most formats, including LTO and TS1140, can read data from tapes written with technology two generations earlier.

Analyzing Disaster Threats

In terms of the type of disasters that may occur, only 3 percent of all disasters are significant natural disasters. Although these may get the most press, they do not form the most pressing threats to data. The most common threat to data is hardware malfunction, followed by human error, software corruption, and computer viruses. Disk is vulnerable to some degree as shown in long-term studies of disk drive failures. (See above graphic.)

Protecting data from logical corruption is one of the primary uses of tape in disaster recovery, and something that disk has never really addressed. If vicious software or viral malware hits a disk, it can spread to the initial disk, the back-up disk, and RAID, which serves as the backup for the backup. It is possible that a disk-only back-up strategy can actually worsen the initial disaster situation against which it is supposed to protect.

The combination of disk and tape for disaster recovery is the strongest defense against risk. The University of Chicago’s tape investment paid for itself with a single incidence of disk failure—without tape, the data would have been permanently lost. Online providers clearly trust tape, as illustrated by Google’s recent Gmail issue. On Feb. 28, 2011, Google posted on its Gmail platform an apology to the 0.02 percent (estimated approximately 150,000) of users who could not access their e-mail. The culprit? A software bug that attacked e-mail in the disk arrays and disk backups across data centers. All copies of the e-mail were unavailable, save those written to tape. The announcement states, “To protect your information from these unusual bugs, we also back it up to tape. Since the tapes are offline, they’re protected from such software bugs.” This succinctly illustrates the importance of using tape for disaster recovery.

Best in Data Recovery Practices: Both Disk and Tape

Although much of this data counters prevailing opinion, the facts are clear that tape, when used properly with disk, offers advantages that disk alone cannot match. Given the improvements in tape and tape automation, it is simply prudent to include tape as part of a disaster plan. Tape helps to ensure data integrity, affordability, and is a smart companion to disk in long-term data protection.

Molly Rector is chief marketing officer for Spectra Logic. Rector brings more than a decade of data storage industry experience to this role, where she helps define and execute the company’s future product roadmap and overall corporate direction, and serves as its primary corporate spokesperson.

Beth Walker is a chief writer for Spectra Logic. Walker has worked at Spectra Logic for more than 15 years, writing about products and the industry, and authoring multiple white papers for the company.