The 10-person IT staff at the New York City Department of Sanitation’s (DSNY) Bureau of Motor Equipment plays a key role in the smooth running of the nearly 6,000 collection trucks, mechanical street sweepers, passenger cars, salt spreaders and other vehicles that make up the department’s fleet.
A loss of critical data at the bureau can have fast and far-reaching implications. “There are a variety of data that we have to maintain constantly in case of snow emergency,” says Marc Williams, the bureau’s supervisor of network operations. “If there’s a snow emergency and a collection truck with a plow attached or a salt spreader breaks down, it’s over — the mayor doesn’t want to know that.”
To more efficiently protect against data loss, DSNY’s Bureau of Motor Equipment in 2004 became one of 30 early-adopter organizations to implement a private beta of Microsoft System Center Data Protection Manager (DPM) — created by Microsoft to provide a low-cost, near-continuous, disk-based backup and recovery solution for the Microsoft Windows Server System. [The product was initially introduced in September 2004 under the codename Microsoft Data Protection Server.] Today, Microsoft’s foray into the disk-based backup and recovery industry reached another milestone with the release of the DPM public beta.
“Our whole goal with DPM is to shrink the operational costs associated with IT professionals having to manually recover lost data and manage cumbersome backup and recovery processes,” says Ben Matheson, group product manager for DPM at Microsoft. “From what our early-adopter customers are telling us, DPM is doing that very effectively.”
The announcement of the DPM public beta coincides with the Storage Networking World conference — a leading conference for IT managers, storage architects and infrastructure professionals — taking place in Phoenix, Ariz., April 12–15. The public beta will be accompanied by two related releases — a software developer kit (SDK) to help Microsoft’s storage partners develop software that can be used to archive data from DPM; and a Microsoft Operations Manager 2005 (MOM) pack designed to facilitate DPM management. General availability of DPM is slated for late 2005.
The Challenges of Data Protection
Data protection is the process of backing up and archiving data for later recovery in the event of accidental deletion or a disaster. Currently, the most common way to protect servers from data loss is to back up server data to removable tapes. Disaster recovery plans consist mainly of manually transporting the tapes to an offsite storage facility.
But tape has its drawbacks, according to Ray Paquet of Stamford, Conn.-based Gartner, Inc., which provides research and analysis on the global IT industry. “Restoring data from tape can be expensive, unreliable and time consuming, often involving manual intervention by IT staff,” says Paquet, who is vice president distinguished analyst at Gartner. “Some organizations are finding that disk-based backup offers a much more rapid and reliable way to handle recoveries.”
Disk-based backup also provides a solution to a key challenge facing IT professionals — the so-called shrinking backup window that results from organizations’ need to back up and retain increasing quantities of data in less and less time. The former stems largely from the relentless digitization of business documents and processes. The latter has to do with organizations’ growing need to maintain round-the-clock system availability.
Customers report that DPM offers key benefits that help them meet these challenges. “It’s going to save a tremendous amount of staff time doing recovery,” says Dan Warren, one of five network specialists on the 15-person IT staff of Des Moines Public Schools in Iowa — an early adopter that is testing DPM on a small number of its approximately 100 servers. “My goal is to put all 60 of the district’s schools on DPM,”
Turbo Charged Backup Speeds
Warren and his colleagues face a shrinking backup window and regularly need to recover lost or deleted files — characteristics that put Des Moines Public Schools at the sweet spot for DPM. They say they are seeing significant growth in the amount of data they have to back up, with the district’s 5,000 employees storing personal documents and grade books on computers, and more than half of the 32,000 students using home directories. “Right now, we’re running out of time to back it all up — there’s not enough hours in the day to do it,” Warren says.
He says the IT administrators do full backups over each weekend — a process that takes about 36 hours using their existing solution for the district’s 16 secondary schools running Windows Server 2003 file servers. When all those servers are protected by DPM, Warren says, the same task will take just 2 hours.
In New York, DSNY’s Bureau of Motor Equipment also saw a huge increase in backup speed using DPM. A 23GB backup using the bureau’s previous solution would take about 48 hours, typically done over a weekend. “And I’d come in Monday and it was still backing up sometimes,” Williams says. The DPM solution does the same job in about 10 minutes — and does it more reliably. “Our previous solution would come up with these crazy errors, partial backups and so on,” Williams says.
The increased backup speed of DPM is due in part to its much more efficient backup method. DPM moves only the byte-level changes of the files on production servers instead of doing a full backup of any file that is changed. To illustrate, if a user has a 10MB Microsoft PowerPoint file and changes only one slide, the tape backup software backs up the entire 10MB document with every change rather than simply backing up the incremental changes, resulting in massive backup and storage inefficiencies.
In addition, the near-continuous data protection offered by DPM allows IT managers to protect file servers more than once per day, even hourly, thus limiting the amount of data a business could lose. That’s an important feature for Des Moines Public Schools, where the 42 elementary schools — which are not yet protected by DPM — have no disaster recovery plan. “If something happened to the file servers, we would lose absolutely everything,” Warren says.
Faster Recovery and More Recovery Options
Customers cite fast recovery of lost or deleted data as the single biggest benefit of disk-based solutions over tape-based solutions. About 80 percent of recoveries stem from users accidentally deleting or overwriting files; the latter frequently occurs when users make changes to a file — a PowerPoint presentation, for instance — then use the “Save” command when they meant to use the “Save As” command.
With DPM, recovering data is as fast and simple as browsing a set of folders on a file share and copying them directly from the server running DPM to the production server — a process that takes seconds or minutes. By contrast, Matheson says, recovering a single file from a tape-based backup system typically takes a couple of hours, and recovering multiple files can take days.
Furthermore, DPM allows for a number of practical recovery options, such as delegation of the recovery process to end users. Traditionally, the only people that can do recovery are IT professionals with access to the tape backup software and tape libraries. But with DPM, end users who have been granted permission to restore deleted or overwritten files can do so themselves from the Tools menu within Microsoft Office 2003 or from directly within the Windows Explorer in Windows XP.
That’s an important benefit to organizations with small IT staffs trying to do more with less. At Des Moines Public Schools, recovering files — especially teachers’ grade books — was almost a full-time job for one IT person. Now, with DPM and minimal training, the teachers are learning to recover their own grade books. “This is just simple — you show them one time, and that’s it,” Warren says.
Williams, at DSNY’s Bureau of Motor Equipment, agrees: “The one thing that’s always burdening a network is employees saying, ‘My spreadsheet is gone. I know I deleted it, but can you get it back?'” With DPM, he says, end users are empowered and IT administrators can focus on higher-priority tasks, thereby lowering total cost of ownership (TCO).
Lower Total Cost of Ownership
A number of other factors contribute to lower TCO. For example, many of the disk-based backup solutions currently available to customers are cost-prohibitive, proprietary hardware solutions. DPM uses industry-standard hardware and software, so many customers have found it comparatively less costly to acquire.
In addition, integration with the Microsoft Windows Server System environment and commonly used business productivity tools such as Microsoft Windows XP and Microsoft Office 2003 means that many customers can take advantage of their existing investments and lower training costs. For instance, DPM contains an interface built upon the Microsoft Management Console (MMC) that many IT administrators are already familiar with. And through integration with the Active Directory directory service in the Windows Server System, DPM auto-discovers all production servers in the environment, and continues to discover new servers added to the network after deployment. “I don’t think it could be any easier,” Warren says. “It’s perfect for a school district.”
Matheson notes that common tasks in many competing backup products are needlessly complex and involve a steep learning curve. “We’ve made DPM so it’s drop dead simple,” he says. “It’s very easy to install, to discover all the servers in your environment, to schedule and monitor and view all the common tasks that an IT administrator wants to do.”
It’s important to note that Microsoft is not recommending that customers rip out their tape-based solutions and replace them with DPM. Matheson notes that tape does have certain benefits — tapes are more easily transportable than server boxes, for example, and they can last for several years. However, the benefits of backing up to disk are compelling. Matheson cites in-house research by Microsoft that suggests 70 percent of the TCO over a three-year period correlates directly with the manual labor of the IT staff. One easy way to reduce management costs with DPM is to centralize the backup of remote or branch offices. At Microsoft, for example, DPM is expected to save an estimated US$800,000 over two years by centralizing the backup of more than 100 branch offices over the wide-area network (WAN) to Microsoft headquarters, Matheson says.
With that in mind, Microsoft recommends using both disk and tape in a three-phase disk-to-disk-to-tape scenario. Under this scenario, replicated data is stored to disk for about 30 days for fast recovery if needed, then backed up to tape for longer-term archiving and disaster recovery scenarios on a less frequent basis. This policy will result in fewer tape backups and less money spent on tape media. “We designed DPM to be complementary to tape backup systems,” Matheson says. “What makes it attractive from a cost standpoint is that you don’t have to rip out what you’ve got.”
Customers are not the only ones who stand to benefit from a disk-to-disk-to-tape scenario. Original equipment manufacturers (OEMs) will have an opportunity to sell more large storage boxes for disk backup, while independent hardware vendors (IHVs) will still be able to sell backup tapes, Matheson says. Meanwhile, Microsoft is taking steps to ensure that the DPM platform interoperates with the tape backup solutions provided by independent software vendors (ISVs). Indeed, Matheson says, Microsoft is relying on third-party ISVs to backup DPM to tape themselves — hence the release of the SDK with the DPM public beta. “It’s another platform for the tape software ISVs to take advantage of and drive incremental revenue,” Matheson says. Some industry partners already have developed white papers on disk-to-disk-to-tape solutions supporting DPM. To download these white papers go to www.microsoft.com/dpm
The Future Vision for DPM
DPM is designed to work exclusively with the Windows Server System family of products. Says Matheson, “From our standpoint, we’re interested in cleaning up our own house first and making sure Microsoft products are protected and supported in the best possible way.”
While this first version of DPM provides support only for Windows file servers, future releases of DPM will support the entire Windows Server System, including Microsoft Exchange Server, SQL Server and others. “In future versions, we could have a heterogeneous story,” Matheson says. “But that decision is going to be based on customer feedback.”
For now, the feedback suggests customers are very supportive of the direction Microsoft is taking with the product. For DSNY’s Bureau of Motor Equipment and Des Moines Public Schools, both Windows customers, the sooner they can protect all their servers on DPM, the better.
“I’m quite anxious to see the public beta,” says Warren. “We put this last release on and saw some of the changes that they made and it just gets better and easier to use.”
Software Development Kit, Management Pack Released
With the April 13 release of the public beta of Microsoft System Center Data Protection Manager (formerly codenamed Data Protection Server), Microsoft is also releasing an SDK and a MOM pack.
Volume Shadow Copy Services Writer SDK 1.0 for DPM: The Volume Shadow Copy Services (VSS) Writer SDK 1.0 for DPM contains sample code that demonstrates methods to easily back up and restore DPM replicas and configuration databases using VSS technology. This SDK is intended for use by Microsoft’s storage partners to accelerate their development of VSS-based software that can be used to archive data from DPM. It implements parts of the VSS standard that are relevant to DPM and is in no way intended to be a replacement for the VSS SDK distribution.
To get the VSS SDK for DPM go to www.microsoft.com/dpm
Microsoft Operations Manager 2005 (MOM): Microsoft DPM 2006 Management Pack enables an administrator to centrally monitor data protection, state, health and performance of multiple DPM computers through Microsoft Operations Manager 2005 (MOM). From the MOM Operator console, the administrator can monitor DPM and network infrastructure simultaneously, analyzing data protection failures in the context of other network failures. From the same console, the administrator can also monitor other mission-critical applications, such as Microsoft SQL Server and Microsoft Exchange. Features include the ability to:
Centrally monitor the health and status of data protection and critical performance indicators of multiple DPM computers
Monitor actionable alerts relating to replica creation, synchronization and shadow copies on DPM servers. The DPM Management Pack filters out alerts that do not require an action, such as a synchronization job that is in progress.
Connect to the protected server that generated a data protection alert in order to diagnose and resolve an issue
Customize views of DPM alerts, state and performance data by creating rules in the MOM Administrator console
MOM pack is available in the CD distribution or download of DPM found on www.microsoft.com/dpm