In corporate environments, file servers are the workhorses that store the majority of employee data. But because they usually get backed up only once a day, if the file server goes down, or a file is corrupted or accidentally deleted, work comes to a grinding halt.
Bringing file protection up a notch is the latest goal of many vendors--from household names like IBM and Microsoft, to startups like San Francisco-based Lasso Logic.
Last month, one year after it was initially announced, Microsoft finally made available its Data Protection Manager (DPM), disk-based backup software that creates several point-in-time images per day of a Windows-based file server.
As a server application that runs on top of Windows 2003, DPM can protect file servers running Windows 2000, Windows 2003 or Windows Storage Server, Microsoft's NAS operating system. As a first step, DPM quiesces Microsoft's Active Directory and deploys agents to file servers or shares it finds in the environment. Those are then replicated to the DPM server. The agents then keep a byte-level log of changes (writes) made to the protected file servers, which are replicated back to the DPM server as frequently as once an hour. Once on the DPM server, the changes are applied and DPM takes a snapshot of the system. That snapshot will act as the basis from which users and administrators can retrieve lost or corrupted files.
On the recovery side, Microsoft's goal with DPM is to allow end users--not only administrators--to recover their own files. Today, that's possible if the file being restored was created using an Office application (e.g., Word or PowerPoint) because Office will allow users to browse previous versions. Alternately, if the end user is running XP Service Pack 2, "shadow" copies of a file stored on a DPM server are visible directly through the XP shell, says Ben Matheson, Microsoft's group product manager for DPM.
Hewlett-Packard was the first vendor to offer a product based on DPM, which it will bundle with two of its Windows Storage Server-based NAS products: the ProLiant DL380 G4 and DL100 G2 Data Protection Storage Server. With a list price of $950 for a DPM server license and three agents, it "provides near-continuous protection for Windows servers at a low TCO," says Duncan Campbell, VP of marketing for HP's StorageWorks Division.
Some view DPM as high-end data protection for the masses. "We think this will enable customers in the midrange who don't have the pocketbook for these high-end vendors and exotic solutions to get 85% of the protection at 25% of the cost," says John Joseph, VP of marketing at EqualLogic, a clustered IP SAN vendor whose arrays have been certified to work with DPM. The remaining 15% of functionality DPM can't provide revolves around synchronous data protection--"which sounds great in theory, but very few people are willing to pay for it," he notes.
However, how many files end users find on DPM depends on where they store those files, as DPM protects only files stored on a file server and not those stored locally on end-users' machines.
Backing up data stored on desktops or laptops "is clearly important for the small- to medium[-sized] business because they store a critical amount of their data on the desktop," says Anna Yen. Yen is co-founder and general manager of Lasso Logic, which this month announced version two of its Lasso CDP appliance, a "true CDP" product that protects files on servers, desktops and laptops as soon as they're saved.
Another product that touts desktop and laptop file protection is IBM's new Tivoli Continuous Data Protection (CDP) for Files.
"There exists a class of data that needs to be backed up immediately," says Chris Stakutis, CTO and creator of IBM Tivoli CDP for Files. Every time a file on a protected server, desktop or laptop is changed, Tivoli CDP for Files saves a copy to a local cache. The contents of the cache are periodically flushed out to the CDP for Files server. Designed with the remote office in mind, CDP for Files cuts down on network bandwidth by using what IBM calls "adaptive differencing," which, according to Stakutis, sends only what's changed over the wire, rather than the whole file.
For Arun Taneja, senior analyst and founder of the Taneja Group, Hopkinton, MA, all this CDP talk begs the question: Do users really need true CDP or is near-CDP enough?
"The situation for file storage has been so deplorable for so long," Taneja says, that most likely, just having access to several recovery points per day instead of one "will be seen as a panacea."
Dig deeper on Tiered storage