RealTime IT News

The State of CDP

Continuous Data Protection (CDP), as the cutting edge of always essential but normally staid backup technology, had a big '05, witness the entry of EMC, HP, IBM, Microsoft and Symantec into a space previously populated by startups, a string of funding and product announcements from those startups, and the establishment of a new SNIA CDP special interest group (SIG).

This year promises to be just as interesting, with products from the major players both increasing the visibility of CDP and making the technology more approachable for enterprise IT.

There are a variety of approaches to CDP, and, as fits a term with such cachet, it is applied to radically different products. It's too early to pick a winner from among these approaches, but it's not too early to spot a trend toward increasing integration of CDP, both with protected applications and with other storage technologies.

A Big Umbrella

Defining CDP is tricky, mostly because the inclusiveness of the term is the source of some contention in the industry. In dispute is the "continuous" component — is the ability to recover from arbitrary points in the continuum a defining characteristic of CDP?

SNIA provides the following definition:

"Continuous data protection (CDP) is a methodology that continuously captures or tracks data modifications and stores changes independent of the primary data, enabling recovery points from any point in the past. CDP systems may be block-, file- or application-based and can provide fine granularities of restorable objects to infinitely variable recovery points."
"Infinitely variable recovery points" would seem to require no fixed recovery points.

But the term CDP has also been used more broadly to refer both to products that allow restoration at the granularity of single write operations and to products that offer restoration only from specific points in time.

The two most prominent examples of the latter approach are Symantec's Backup Exec 10d and Microsoft's Data Protection Manager (DPM), which provides what Microsoft refers to as "near-continuous data protection." Each of these record and replicate changes continuously, but snapshot data on the server only at one-hour intervals, so historic data is available only at these increments.

"I think customers are a bit confused over the difference between a lot of snapshots and continuous data protection," says Enterprise Strategy Group analyst Brian Babineau. Babineau says both approaches have a place: snapshot and replication for less mission-critical applications that can tolerate a small window of data loss, and true CDP for applications where a zero or near-zero RPO is required.

Part of the confusion arises from trying to group these two very different segments under the same umbrella. Replication and snapshot solutions do not provide arbitrary recovery points, but they are aimed at solving general backup and recovery problems at the departmental and SMB level. In contrast, true CDP aims to provide zero data loss for targeted critical applications in the enterprise.

"There are those that are saying you have to be able to dial back to any arbitrary point in time," says Michael Parker, group product marketing manager at Symantec. "From our discussions with the mid-market, that's not exactly something they are looking for."

In the mid-market, CDP products provide automated, regular backups and can be part of a shift to an overall disk-to-disk-to-tape strategy. For these customers, says Parker, "The nature of data protection is changing. People are rethinking how they do backup."

Mixed Success

Despite all the buzz, deployments are limited, at least at the enterprise level. "People are just starting to get their toes wet in this technology," says Babineau. "They haven't deployed it full-fledged across the entire application infrastructure."

Mendocino Software is one of the startups that has seen some success with CDP, with its block-level CDP software being resold by EMC and HP. Eric Burgener, vice president of marketing at Mendocino, agrees that the market is in its early stages, at least in the enterprise.

"There are not a lot of deployments going on right now," says Burgener. "There are a lot of trials going on in the enterprise arena. More of it has been deployed down in the lower end of the market."

Like Mendocino, Revivio sells block-level CDP software. But Revivio senior vice president Kirby Wadsworth says his company is starting to see the end of the education phase and a move toward purchasing. "What we're seeing now is a pretty brisk uptake in customer adoption," says Wadsworth.

For application-specific CDP products, there's less of a question of awareness of the technology, as customers already familiar with the problems of backing up Exchange, for example, seek new solutions.

"I have never heard a customer say, 'I want CDP,'" says T.M. Ravi, president and CEO of Mimosa Systems. Mimosa uses CDP technology to solve Exchange backup problems, and leverages the backup data to provide additional features, like searches conducted against historical data.

Enter the Majors

2005 saw the introduction of file-based CDP software from IBM, block-based CDP offerings from HP and EMC, and snapshot and replication tools from Symantec and Microsoft. Analysts and vendors agree that the entrance of these names give CDP more traction.

"It's sort of a blessing of the concept of CDP. When the likes of EMC and HP are reselling it, it means that their customers are asking for it," says W. Curtis Preston, Vice President of GlassHouse Technologies.

EMC licenses Mendocino's CDP technology, but the company builds a full recovery platform around it by further adding management and application integration.

"What we do," says Rob Emsley, director of product marketing for EMC Software, "is integrate the RecoverPoint engine and the RecoverPoint host drivers with a product called Replication Manager, which provides the management interface and the application integration." Application integration adds significant value, so it's not just a matter of EMC re-branding Mendocino's products.

"The majors are all taking positions," says Revivio's Wadsworth. "You'll see announcements from all the major storage vendors in early 2006. By the end of 2006, there'll be firefights all over the place, because all of the major OEMs will be competing aggressively against each other for this business, because it's new, it's exciting, it's high value-add. There'll be quite a firestorm by the end of the year."

Babineau thinks the startups that have been longer in the market will benefit from the move. "They will certainly draft from the likes of EMC creating market awareness," he says.

But Mendocino's Burgener has a different outlook. He sees the bigger vendors displacing the startups in the field. Burgener's position is understandable, since his company long ago took an OEM-only tack in recognition of the difficulty of selling critical infrastructure products directly to large customers. Naming EMC, HP, IBM, Sun, HDS, Microsoft, Veritas, Cisco and Brocade, five of which have made recent CDP moves, Burgener says, "Fortune 1000 folks like to buy their enterprise storage infrastructure solutions from one of those nine guys."

Toward Greater Integration

The big names may also help drive an emerging trend in CDP, the trend toward deepening application integration with low-level, block-oriented CDP technologies.

"One of the catalysts that will assist in that is the Mendocino-EMC relationship," says Preston. "Hopefully, the giant gorilla that is EMC can help put some pressure on the application vendors to work with the concept of CDP. Up to this point they really haven't."

Mendocino has already disclosed that application-specific plug-ins are part of its plans for 2006.

From the application side, Ravi thinks that even tighter integration than that planned by block CDP vendors is essential. "CDP solutions for applications have to be not just aware of the application, but also intelligent in integrating with the application," says Ravi.

Ravi says one of Mimosa's strengths is its ability to restore application objects (a mailbox, for example) to deal with less-than-catastrophic failures. "90% of the failure situations require the restoration of fine-grained application objects," says Ravi.

For its part, Mimosa plans to push its CDP coverage beyond Exchange in 2006 to include external documents.

Wadsworth says one of Revivio's targets is greater integration with ILM products.

"Towards the end of 2006 and into 2007, you're going to start to see integrated ILM and CDP solutions, where CDP literally feeds the ILM infrastructure," Wadsworth says. "Everything flowing through a CDP stack makes it really easy to apply policies to the data without impacting the online data flow. And that's where I think you really start to see the power of this. Not just in the instantaneous recovery or the DR components of it, but the long-term ILM integration."

What about further into the future? Burgener sees a scenario where the technology is commonplace, but only for the most critical applications. "If this whole CDP thing is wildly successful from an industry point of view," he says, "what you might see two or three years down the line is maybe 10-15% of a company's applications might be leveraging CDP."

Ravi thinks the data-capture end of the technology is bound to end up as part of the infrastructure. There are two likely places for CDP data capture in the long term, says Ravi. "One is sitting in the operating system itself, coming from a vendor like Microsoft," he says. "The second is sitting in a disk subsystem, sitting in an array controller or a networked array controller from a vendor like Hitachi or EMC."

In the end, the future of CDP may be all about this kind of integration, as part of the infrastructure or at least as part of a strategic backup offering.

"Continuous protection should be a feature of any viable data-protection solution," says Symantec's Parker. "A number of these startups out there either, frankly, will go away or be acquired, because that kind of functionality will be coming to a larger data-protection solution."

EMC's Emsley has a similar take: "One of the challenges that some of the offerings in the CDP market have is that their relationship to a customer's existing infrastructure is not very tight. The vendors that offer CDP as an integrated component of their overall recovery offerings are going to be more successful."

For more storage features, visit Enterprise Storage Forum Special Reports