November 2010

Smart Practices

By Jennifer Zaino

Many businesses have considered cloud options to help get data back online when disaster strikes, but they may not be aware of many more ways the cloud can benefit business continuity.

The University of Pennsylvania’s Wharton School of Business, for example, mixes established and new technologies for data storage and recovery needs. “We have fairly good, well-tested disaster recovery in the more traditional sense,” says CIO Deirdre Woods of an environment that includes some 500 virtual machines hosting data for 5,000 students, 900 faculty and staff every day. Applications range from e-mail to research services. Its “old school” practice is to back up some 35 terabytes of data nightly to tape that it ships to an off-site location, Woods says. But since she knows that virtualization and the cloud can add risk when it comes to recovering data, she is broadening her horizons. In addition to its tape backup processes, Wharton just signed a contract with a service provider to move data into the cloud for disaster recovery purposes.

As a pilot project, Wharton is replicating the data from its main departmental file server; when users in Philadelphia make a change to a document or spreadsheet, it is immediately copied to the remote service. Woods hopes to expand this offering to the rest of the school's data repositories, as well as its Web sites. The cost, she says, is around $2 per gigabyte, including setup and transfer charges.

Considering the Options
In fact, there are multiple ways to look at the cloud from a business-continuity perspective beyond traditional data storage, Woods says. Despite a poor experience with a hosted e-mail provider in the past that led Wharton to bring regular messaging services back in-house, the institution is again relying on a cloud provider to step in, if necessary, to deliver dial-tone service, message flow and mailbox access recovery services. E-mail “is the first thing to get back up in any crisis,” says Woods. “I need to provision e-mail somehow if I have no e-mail on site,” she says.

Mark Bowker, senior analyst at Enterprise Strategy Group (ESG), has seen some interesting twists on business continuity as a result of cloud services. In one case, a hedge fund manager was concerned that an event could keep employees from entering the building — which literally would result in the company’s losing money — so desktops and applications were moved off-site to a distant hosted environment. The company also replicated some primary apps to a secondary data center in case it was needed. “If there were a local outage in the building, they’d still be able to access their desktops and applications,” Bowker says. “They would remain productive and be able to serve clients from home or some alternative location. That’s a great way to stay up and running — leveraging resources you don’t own.”

When the Cloud Is the Business
For other organizations, the cloud is taking on much more central importance as the means to a redundant IT infrastructure that protects customer-facing offerings as well core business operations. OpenAmplify, a provider of semantic software solutions, is far along the curve when it comes to cloud services. The Web tech company moved from a classic co-location server farm with some 110 systems to the Amazon Elastic Compute Cloud (EC2) and S3 Simple Storage Service a few years ago.

With operations in the U.S., the U.K. and Sweden, OpenAmplify originally decided to leverage the cloud to preempt business risk associated with scaling its dispersed operations — something that many businesses are still thinking about. “We wanted to architect our [platform] to run hundreds of millions of transactions a day,” says co-founder and CTO Mike Petit. “But we didn’t know how quickly people would need us to do that, and we didn’t know how to build something with the potential to scale to that size and adequately test on a small number of machines,” he says. “The cloud shifted the risk” of miscalculating scale.

OpenAmplify has found that the ability to scale resources to support customers is a form of business continuity in itself. In addition, the international nature of its operations enabled the company to turn to a multiple-zone strategy on the EC2 cloud which the company uses for everyday service functions. For instance, performance will likely be better for a European customer if the client can access the software platform from an EC2 machine located somewhere on the continent. But the payoff in operational continuity, Petit says, is that “there are certain data sets I can’t afford to have offline, so staging them out to physically diverse locations is a nice strategy.”

To ensure business continuity, those data sets, which include product code, are synchronized on three discrete physical servers on the cloud in different availability zones. They are also backed up on two discrete physical servers in the U.S. and on hard media stored off-site.

Considering that Petit has never lost a production server or data on the cloud, having so many layers of protection might be viewed as overly cautious. Nevertheless, he speaks to many CIO’s concerns when he says: “It’s one thing to lose an instance and have to start up again,” he says, [but] “it’s another thing to lose data.” Amazon is keenly aware of that, he says, but still, Petit doesn’t believe he can be overly cautious when it comes to the software code that’s behind its core product: “Never lose the code. It’s where the IP [intellectual property] and investment are].”

He follows the same strategy with other important data sets, such as customer API keys — all backed up in as many physical, nonphysical and geographic ways as OpenAmplify can devise. That includes replicating cloud-based server databases through the Internet to noncloud-based servers. 
There’s only one area where OpenAmplify takes a contrary approach to using the cloud as its primary operating model: It runs e-mail on an on-site mail server in its Stockholm office, and in this case, the cloud is called upon only to back it up.

Gaining Comfort
Not everyone is as comfortable as OpenAmplify in its reliance on the cloud, and some still have concerns about it as a disaster-recovery outlet. Even among business-savvy academics at Wharton, there are challenges to overcome in terms of data stewardship and governance, Woods notes.

Nevertheless, adoption will gradually follow, Bowker says. Given that in many organizations backup and disaster recovery “fall a little off the priority list all the time,” the cloud could be a way to ensure that it gets done. “You can leverage some third-party provider, or cloud service to do that, and take it off the list,” he says.

That said, organizations may need assurances that their data will be backed up only to geographically nearby servers or meet other industry or government regulatory requirements. Another important consideration, Bowker says, is what happens in two or five years when you want to change the providers. “What does it take to change the history of your backup and move it and the log files associated to a different platform?” Getting that straight with your cloud service provider before you sign on the dotted line can save you a lot of headaches down the road.

Jennifer Zaino is a New York-based business/technology writer and editor.

ASK THE EXPERT

Deirdre Woods, CIO, University of Pennsylvania, Wharton School of Business

Deirdre is Associate Dean and CIO at Wharton where she leads the IT team in delivering technology resources to faculty, students, staff and alumni. The team develops strategies to “engage 21st-Century learners.” Deirdre is also responsible for Wharton Research Data Services (WRDS), a computing platform, data warehouse and scholarly research commons used by more than 275 universities worldwide.
During more than 20 years at Penn, she has served as Senior Director for Wharton Computer and Information Technology and as Director of Faculty Services and Networking. She earned a degree in economics from St. Lawrence University and a Master’s degree from Duke University.

Mike Petit, Co founder, CIO OpenAmplify

Mike has over 30 years of professional experience integrating the technology and marketing disciplines at companies including Ogilvy Advertising and Scudder. A co-founder of OpenAmplify’s parent company, Hapax Ltd., Mike has served as CIO since the company started. He is the inventor of the OpenAmplify service and is a graduate of Vassar College.

Mark Bowker, Senior Analyst, Enterprise Strategy Group

Mark is a senior analyst at ESG focusing on virtualization and cloud computing. He researches the various virtualization technologies available and the impact these solutions have on IT strategies and the broader marketplace. His other research areas include cloud computing, data center management, and application workload deployment in next-generation data centers, as well as the external influences that drive the adoption of these technologies.
Previously, Mark ran the IT organization for a business consulting and technology services company and is a Microsoft Certified Systems Engineer. He has experience in designing, implementing, and expanding network and system infrastructure for global organizations. Mark holds a degree in Management Information Systems from American International College and is certified in Unix administration and C++ programming. He is also a private pilot.