Making the Case for Both Disk and Tape

Until recently, hard disks weren't practical for backup, mainly because the necessary drives and RAID controllers were pricey. But D2D storage is affordable because RAID configurations are now being built on the inexpensive, high capacity ATA/IDE technology. Silicon for an ATA RAID controller today, for instance, costs less than silicon for SCSI (see costs chart, below).

Reel Hard Choices

There are advantages and trade-offs to both disk and tape technologies. Tape remains the dominant technology for long-term storage of archival data because of its ample capacity, portability and life expectancy.

With hard disk drives, capacity is fixed, so if you want to double your RAID array's capacity, for instance, you have to buy another RAID enclosure with an equivalent number of drives or buy higher-capacity drives. A tape cartridge, on the other hand, instantly and inexpensively adds capacity: With each tape you buy, you can double your tape-drive capacity at a fraction of the drive cost.

Tape's biggest limitation, of course, is its large backup window. Tape is sluggish because of its linear nature and slow transfer rates. It takes longer, for instance, to access data in the middle or at the end of the tape. But if speed is less of a concern than capacity for your organization, go with tape. Be aware, though, that the more backup tapes you fill, the more you'll have to manage.Portability-wise, tape is handy because you can easily store your data off-site by moving tapes to a secure location. That's ideal for disaster-recovery planning, for which off-site data vaulting is crucial. Still, portability has its risks. Tapes and, consequently, your corporate information can be stolen at or en route to an off-site storage facility.

Heftier disk systems obviously aren't as easily transported to an off-site location for disaster recovery. But one way to use disk technology for disaster recovery--and this works only if your data doesn't change much--is by mirroring pairs of disk arrays over a WAN link. This isn't always practical, however, given the cost of bandwidth and the relatively low speed of many corporate WAN links.

When it comes to storage speed, disk wins hands down for backing up and restoring your data. D2D has a higher data rate than tape, generally at least two to three times that of the newest tape drives (not taking into account linear access and media-automation access times). It accesses data randomly, which is more efficient than linear access. It restores data with standard file tools, which have obvious ease-of-use advantages over, say, tape backup software.

One D2D product that takes a unique approach to supporting both disk and tape is Quantum Corp.'s DX30 disk-based backup device. The DX30 has a tape interface to the SAN with standard backup and restore software as if it were a tape device, but it operates at disk speed.

Taking point-in-time data snapshots is a big plus for disk as well. If you need to update a critical database, for example, a D2D device can take a snapshot, and if you need to back out of the upgrade, the device can quickly restore your data according to the snapshotThese benefits D2D are significant, but before you're sold on the technology, note one trade-off: Going with D2D means more disks for the storage administrator to manage because mainline storage-management software doesn't handle the cheaper, specialized D2D storage.

Deciding which medium to use when and where depends on your business and operations. One good way to blend tape and disk technologies is for backup and restore functions. A sales database, which is valuable and constantly changing, is a prime candidate for near-line backup storage with a D2D system. It would be backed up from the main-line storage disk to the near-line D2D array quickly, and could be restored fast, too. From there, sales data could flow to a slower tape medium for offline storage.

Another way to mix and match tape and disk is by using archival backups and optimizing your main-line storage. With the archival approach, determining the backup tiers requires that you first carefully examine and classify the data you're backing up. Internal memos, for instance, are first stored in your main-line devices, backed up to a D2D device and, finally, to tape for off-site storage. When this data ages and is less frequently accessed, you can move it off your main-line storage altogether, and then down to your D2D device. From there, you can transfer it onto tape only or to a long-term archival storage system, such as an optical WORM (write-once, read-many).

Moving data from one storage tier to another as it ages ensures that the most frequently accessed data is on the faster, main-line, disk-based storage, and older data kept for archival purposes is moved to your slower tape storage. So your main-line storage accommodates critical data, and lets less expensive storage take care of the rest.

Window of OpportunityThe process of protecting your company's data encompasses every stage of the data life cycle--creation, storage, backup and, eventually, erasure. You have to decide what data retention times, restore times and backup windows are acceptable for your operations.

Where you set these targets for protecting your organization's data depends on your line of business. Many retail companies, for instance, experience peak times during the day, so they typically prefer evening backups. Business managers should work with IT to determine an organization's target backup windows, restore times and data retention policies to make sure the company gets the most out of its storage architecture.

And while mixing tape and disk is the most efficient way to handle data backup and restoration, the two technologies are not exactly plug and play. The missing piece of the puzzle for integrating disk and tape is integrated disk-and-tape management and backup software. Today, tape software has to be adapted to disk and vice versa, which makes for some disjointed and inefficient tools. The next frontier is true storage management, with software that works for both mediums.

Steven J. Schuchart Jr. covers storage and servers for Network Computing. Previously he worked as a network architect for a general retail firm, a PC and electronics technician, a computer retail store manager, and a freelance disc jockey. Write to him at [email protected].

Post a comment or question on this story.When it comes to desktop and laptop backup for remote users, disk-based technology is the clear winner because of its speed and ease of use. Tape devices for these systems can be terribly slow, which can confuse and frustrate end users.With disk backup, end users can run familiar tools such as Microsoft Explorer and backup software designed for ease of use. Also, external disks are simpler to handle than bulky portable tape drives with their associated fragile media. With the use of speedy USB 2.0 and Apple Computer FireWire peripheral interfaces, connectivity is a snap, too.

But when you're talking corporate servers or other bigger systems, deciding which way to back up isn't so clear-cut. Businesses need off-site storage, even if it means Joe in IT has to take home the tapes. Disk-based backup can work well, for instance, in remote locations where there's no on-site IT support. A backup often fails when the non-IT person tasked with tape-change duties doesn't do the job. Off-site locations can also have less than perfect conditions, like dirty tape drives, that can cause unnecessary backup problems. That's where a D2D system makes more sense.

Still, disk-based near-line backup has its problems, too. You can inadvertently replicate an error or virus on your only backup when you deploy the point-in-time or incremental snapshot features of a D2D device. And the ever-changing nature of the snapshot can leave you without a viable backup. To prevent these problems, back up to a tape device once a day as well, or keep a separate, once-a-day snapshot of the data for up to a year (depending on your company's data retention policy).

ATA (Advanced Technology Attachment) technology has finally shaken its bad reputation as an unreliable interface for storage devices. While SCSI and Fibre Channel drives are built with longevity and reliability in mind, that wasn't always the case for ATA. But the quality of ATA devices has much improved, and ATA and its immediate successor, Serial ATA, are now giving SCSI and Fibre Channel drives a run for their money. Although ATA is still slower than SCSI and Fibre Channel--anywhere from 50 Mbps to 150 Mbps--it's becoming more popular as an inexpensive backup and recovery option.

The speed gap, however, has gradually been narrowing over the past several years. Low-end drives have become faster and their capacity has increased. These low-cost ATA drives with huge capacity led to the development of today's D2D (disk-to-disk) near-line backup devices and an explosion in the NAS (network-attached storage) market. D2D vendors are including interfaces on their arrays to handle direct Fibre Channel connections to the SAN or SCSI connections, so ATA has finally found its way into the data center.