"Very few companies do backup well," says W. Curtis Preston, vice president at GlassHouse Technologies, a storage consultancy in Framingham, Mass. "Most have significant challenges."
Can't buy more time
Driven by an increasingly electronic workforce, and regulations that mandate data retention, corporations back up more data every year. At the same time, those workers are intolerant of any application downtime, preferring their applications to be available 24/7.
"Users assume data will be backed up without impacting the application and they want instant restore," Preston says. "They don't want a four-hour or 12-hour answer."
This leaves storage managers with the unenviable job of trying to squeeze more and more data into an increasingly small window of time for backup. Small wonder so many companies have a hard time doing it well.
"Data protection has always been a huge headache," says Chuck McQuade, a system specialist at the University of New Mexico (UNM) in Albuquerque.
Signs of progress
There are signs, however, that things are finally improving for storage managers who want to craft successful backup strategies. The move toward disk-based backup technology is an important step forward, and companies are melding these new products with management-based tactics that treat backup like the business problem that it is.
Companies can start by making sure that their current setup is operating at top efficiency. For example, companies may find that they are not backing up all the data they need to, or are neglecting to make copies of tapes. Moreover, by evaluating their current backup capabilities, something that can be performed either inhouse or by an outside firm, storage managers may be able to tweak their backup configurations to significantly increase both performance and integrity.
Another smart move is to make sure that the right data is getting the red-carpet backup treatment. This involves knowing what you have in terms of business data, and prioritizing it according to its value to the business. The best measurement is to assign realistic recovery times to each priority level of business data. Not every piece of data is going to need the fully redundant live failover system that a money fund trading system does, for example.
Crying tiers, ILM
With data prioritization, companies can then tier their backup and recovery technologies, much as primary storage is now tiered. Doing so ties in snugly with the storage strategy known as information lifecycle management (ILM), says Mike Karp, a senior analyst at Enterprise Management Associates in Boulder, Colo. "The purpose of ILM is to get maximum value from anything you invest in, because you're not taking up time with low-value data being backed up to a high-end device," he says.
As an example, McQuade has recently installed continuous data protection (CDP) technology from Revivio and plans to use it in conjunction with disk-to-disk-to- tape to tier his data according to business priority. "One of the things that CDP will let us do is know in advance how we will protect certain data based in its business value," he says. "If it's mission critical, we can send it to CDP. If it's more operational, it might go to more traditional backup technology. It's a tiered approach to data protection based on different classes of backup services." McQuade is hopeful that he'll be able to finish building his tiers of data protection over the next 12 to 18 months.
Once the planning is out of the way, storage managers can evaluate technologies that can help them reach their backup goals. Here, disk-based backup becomes an important new technological priority. "The No. 1 strategy for many companies is to stop going directly to tape in any form," Preston says. "Every major client that we have has already migrated to disk as a primary storage mechanism for backup."
In addition to strong planning, here are some technologies that users and experts cited as promising in the quest for better backup.
1. Nearline backup
More a concept than a strict technology, nearline technology is the middle road between online access and offline storage, such as tape. Data that's backed up nearline might be on a serial ATA drive -- something that's slower to access than high-end disk, but infinitely faster than tape.
For example, Bill Polymenakos, director of technology for Ecker Enterprises, a specialty construction company based in Rosemont, Ill., instituted some nearline disk-based backup when he installed an ISCSI SAN with integrated backup technology from StoneFly Networks in San Diego, Calif. "What we wanted to do is improve restore time," he says. "By backing up data nearline for a few weeks, and then making tapes to move offsite for a more historical-type disaster, it allows us quicker time to restore."
2. Virtual tape library
Software that runs tape-based backup is not geared to use native disk drives in their native format. Layering virtual tape software between disk and traditional backup software fools the backup software into thinking that a disk drive is actually tape, which allows companies to install disk-based backup without having to overhaul its current backup infrastructure. "You don't have to throw out an entire investment in backup software, which could be a multimillion investment for a large organization," Gerr says.
Large storage vendors have come out with virtual tape libraries in response. EMC Corp., for example, has a Clariion library based on Falconstor VLT technology. Neartek and Diligent are other emerging vendors in this market, Gerr adds.
3. Continuous data protection
CDP essentially provides a rolling database of every single change made to a company's data, keeping a time stamped record of each change. The result: near instantaneous data recovery, pulled from just about any moment in time. Notes UNM's McQuade, "We'll be able to back up and rapidly recover our mission-critical databases," including a large ERP implementation currently under way.
While promising, the market is still pretty small. "It's a little ahead of its time, but as recovery point objectives become more stringent, more people will evaluate it," Gerr says. "We think that in the second half of next year into 2006, we'll see more active employment of this technology." Vendors include Revivio, Mendocino Software, Alacritus and XOsoft.
4. Data reduction technologies.
Known by terms such as commonality factoring, coalescence or deduplication, these technologies seek to reduce the amount of data that's backed up in the first place, which will serve to speed backup and recovery times of the remaining data. The concept works this way: Say a colleague e-mails out a Powerpoint attachment to 10 coworkers. Most backup systems would back up all 10 instances of the attachment, even though much of the data is redundant. Software that uses data reduction technology, however, saves just the changes that have been made to each instance. So if only one slide was changed, that's what gets backed up out of that instance. "It's had a lot of play in the last four or five months," Karp says. "It can be very compelling—once you do an initial backup, you don't have a lot of stuff to write." Data Domain and Avamar are two vendors in this area.