Extra protection

Is application-specific backup software worth the added cost?

For a long time, data backup has been a relatively simple affair involving quick copies or snapshots of daily data changes and more thorough backups once a week to tape. If a system went down, administrators could retrieve the tapes from wherever they were kept for safekeeping and restore the afflicted system.

General-purpose backup software was usually good enough for most, if not all, of the applications and data types a typical organization might have. Sometimes an application might require more frequent backups or a longer retention period for archival tapes, but generic backup software could still do the job.

However, as agencies increase their dependence on technology for many of their basic operations and often run those systems around-the-clock, some are turning to more specialized software to boost the performance of their backup and archiving routines. For them, the old timeframe of taking hours or even days to restore a system is no longer acceptable.

For example, electronic messaging has always been a mission-critical application at the Marine Corps Network Operations and Security Command, but it could become even more so in the future. So command officials are rethinking their backup routines.

“We are changing our viewpoint on backup and restore,” said Dan Carroll, a senior network architect at Smartronix, a solutions provider hired to keep the Marines’ global network and communications running. “In the past, we’ve looked at backup as a resource in case the [messaging] system goes down and has to be brought back. Now we are trying to go to a position where e-mail never goes down.”

To help reach that goal, the company uses Network Appliance’s SnapManager for Microsoft Exchange because it almost instantaneously backs up and restores files, Carroll said.

SnapManager uses native Exchange backup application program interfaces to create snapshots of e-mail databases that take only seconds to complete, regardless of size. It can take as many as 250 snapshots of any volume on a disk without interfering with business or data flows, said Patrick Rogers, NetApp’s vice president of products and partners.

“We do that in just one read-and-write procedure, transparent to the application,” he said. Like NetApp, other vendors that offer such software also provide optional, application-specific backup clients, or agents. Agents are available for e-mail programs such as Exchange and IBM’s Lotus Notes and for the most popular databases from vendors such as Oracle, Microsoft and others.

Application-specific software gives systems administrators the performance and level of control they need, said Scott Antczak, a storage specialist at reseller CDW. For example, standard backup programs working with Exchange files tend to back up and restore files in broad swaths, which can consume excessive system resources and reduce efficiency.

Administrators prefer unobtrusive backups and the ability to restore an individual user’s e-mail files or even specific messages, Antczak said.

“Because the agent knows such things as where each file is located, an administrator can set very specific rules for what needs to be backed up and when,” Antczak said.

That’s one reason why NASA’s Jet Propulsion Laboratory uses such agents, said Luke Dahl, a service engineer in the lab’s Information Services division. The organization needs to back up and archive some 120 terabytes of data according to its class and criticality.

“We need the ability to set up policies and automate backups. Those are very important functions for us in the backup software we use,” he said. “It also allows us to dictate where backups are stored until we can migrate the data to tape, which is still needed for disaster recovery.”

More intelligent archiving
Dahl considers backup and archiving to be different shades of the same process, with backup a higher priority because it deals with the most recent data. However, Laura DuBois, IDC’s research director of storage software, said the marketplace views the two as separate because of the differences in their requirements.

“Archiving requires such things as content-based indexing of data in order to be able to retrieve it according to the specific time it was created, to be able to meet audit and litigation needs,” she said. “That’s information management, whereas backup is generally considered a part of data management.”

Organizations that have had to deal with a discovery request without good archiving strategies in place have quickly found how painful the process can be, said Nick Mehta, senior director of product management at Symantec. In that situation, responding to a legal request for information can involve teams of people manually sifting through reams of data on backup tapes.

“Each case can cost millions of dollars,” he said. “Those organizations that have not gone through it are quickly being educated by their legal side about the need to be more proactive.”

Symantec’s Veritas Enterprise Vault software provides automated, policy-controlled archiving to online storage. The archived information is available for fast retrieval using built-in search capabilities.

As with backup tools, the archiving software uses agents to target specific applications. With Exchange, for example, Enterprise Vault archives and indexes all messages according to policies set by an administrator then stores them in their original form, including attachments.

But archiving is more than just being able to search by keyword and pull data out of a central repository, said Brian Peasland, a senior database administrator at SGT, a contractor working with the U.S. Geological Survey’s National Center for Earth Resources Observation and Science. An archive management system should preserve the relationships between information so users can retrieve information in its full context.

For example, the center has a help desk that fields user calls and then routes them to analysts for resolution, Peasland said. Call records pile up on production systems, but if he deletes them, he risks not being able to properly archive the projects later.

“I can delete calls, but then the associated tasks may be left dangling, so then at some time I also have to go in and pull those tasks,” he said. “If I can’t archive properly, I lose all of those associations, and the only relief is to delete an entire file, which is obviously not what I want to do.”

Peasland has not found commercial software that can meet all of his requirements, so his team has developed much of the software its uses, he said.

Robert Rosen, chief information officer at the National Institutes of Health’s National Institute of Arthritis and Musculoskeletal and Skin Diseases and president of the independent IBM user group Share, thinks application-specific archiving and backup software might not be on most people’s radar, though he knows from experience that it should be.

“In a previous job, we didn’t have archiving, though we were aware of it,” he said. “Then we got a discovery request, and I had to put three people on it full-time for several weeks. That’s when I decided it was more expensive not to do it.”

Worth the extra expense?

Because it is used only for selected programs, application-specific backup software will cost more than general-purpose backup products, which can have their cost spread across a larger number of application and data types. However, many information technology shops find that the application-specific software will deliver performance benefits in all three facets of data protection, which can help justify the additional expense.

  • Backup — Traditional backup software handles all applications the same way, making little or no distinction in how different programs interact with the operating system, store files or link data elements. By comparison, application-specific backup software understands the way the programs organize and manage data, allowing it to perform more sophisticated tasks. For example, many products run in the background, so applications don’t have to be shut down to back up data, even if files are open and users are working on them.

  • Restore — With traditional backup software, organizations typically back up data only nightly or even less frequently. That means the new data created between backups is vulnerable. The new crop of application-specific backup software can copy data at intervals of an hour or less. The software can do that at the object or message level, so it can restore each data element as needed rather than having to restore the entire volume, substantially cutting the time it takes to restore a downed system.

  • Archiving — The conventional approach is to store backup data on a long-term storage medium such as tape with a limited amount of identifying information about the location of datasets. Archiving is undergoing a revolution after the creation of rules about how to handle data and when to make it available to regulators and other oversight bodies. Numerous application-specific archive products can locate older data with great precision and access it in minutes or even seconds using content-based indexing schemes and disk-based storage, which is faster than tape.
County trashes old backup routine

Minimizing the risk of losing data and speeding system recovery after failures are top goals of many computer-dependent government operations, no matter what kind of work they do. For example, the Integrated Waste Management Department (IWMD) in Orange County, Calif., has compelling business reasons for pursuing such recovery capabilities.

The department must provide continuous data to the fleet of hundreds of solid waste haulers the county hires to move its waste to landfills.

“There are tonnage limits imposed on how much waste the landfill can take each day, so we have to know what each of the trucks are carrying,” said Patrick Copland, a senior technical systems specialist at the department. “That information is forwarded to our headquarters from the landfill site, and we then also have to post it on the Web every 10 minutes so the haulers can know how much they have hauled.”

The private haulers also have a critical business need for that information because the department charges them $27 a ton for bringing the waste to the landfill, and then they have to charge the city to get their money, Copland said.

In addition, regulators and others occasionally ask the IWMD about the landfill, and the department needs to be able to retrieve the data quickly, either from the live database or archive.

“There’s daily pressure on us to run reliably,” Copland said. “We just can’t afford any significant server downtime.”

To provide an operational safety net, the department recently bought a storage system from Compellent Technologies. The system includes the company’s Data Instant Replay data protection software. It provides backup snapshots of the IWMD applications and data every 15 minutes. Recovery time for lost files or subdirectories is less than six minutes, Copland said.

The Compellent solution also backs up the server operating system to prevent potential downtime from failed upgrades. The IWMD previously backed up the servers to tape with regular backup software, but the department found the server recovery process using this routine unacceptably long.

“We had occasions over the last year when we couldn’t use the server because of problems with upgrades,” Copland said. “Now we can pick a time when the operating system is stable and peel it off, then use that to reboot. It means we can bring the server back online almost instantly.”