Panasas clustered NAS and Amazon S3 cloud data storage help e-Learning provider

E-Learning provider Connect for Education uses combination of on-premise clustered NAS and Amazon S3 cloud data storage to optimize performance, cost and data protection.

An e-Learning software designer and hosting provider set up a hybrid cloud data storage architecture with a combination of Panasas Inc. clustered NAS systems and cloud storage through Amazon Simple Storage Service (S3) to handle files and databases.

Connect for Education Inc., a Reston, Va.-based e-Learning firm, replaced direct-attached storage (DAS) and file servers with a high-performance Panasas ActiveStor clustered NAS system in 2008, said CIO Luis Velazquez. The original purpose of the clustered NAS was to support the serving of static content for the e-Learning courses the company designs for college clients.

However, as the deployment progressed, Velazquez said his company quickly began running up against its initial storage capacity on two Panasas clusters, which held 9 TB and 18 TB of capacity and were deployed at both the company's headquarters and its data center at a separate location. As capacity began running out, Velazquez began searching for ways to expand his storage using a tiered model to cut costs.

 Besides content for courses, Connect for Education also hosts databases that track student grades and test results.

"Our database is the critical piece," Velazquez said. "We feel it's vital and central to our business to keep it in the security of our own data center." So besides the need for more storage capacity, the company was concerned with data protection for its Web load balancers and front-end databases, as well as how to deliver static content across the country from its centralized data center.

"Aggregating our four Gigabit-Ethernet links wasn't enough to sustain high-volume throughput across the country," he said. "So for that portion, we decided to use cloud servers, and Panasas became a quick and ready backup."

Today, two Panasas systems -- originally intended for each end of the wire between Connect for Education's business headquarters and data center -- live side-by-side at the data center. They're used to make redundant copies of the front-end critical databases still hosted there. In addition, the data center still handles the systems that perform tests, student assessments and authentication, while storing flat files on Amazon S3. S3 also serves as a content delivery network (CDN) with its network of multiple data centers across the country.

Most enterprises have been reluctant to port more than backup and disaster recovery data to cloud data storage, largely due to perceptions of problems with the cloud's reliability, availability and security. Velazquez said his current architecture offers the best of both worlds. He hosts sensitive systems on-site while taking advantage of cloud data storage flexibility for growing unstructured content.

The Panasas systems also provide redundancy and high-performance restores of critical database snapshots Velazquez makes daily with Symantec Corp.'s Backup Exec System Recovery (BESR) to maintain the availability of Connect for Education's core systems.

"It's a lifesaver, because if one of our servers is down, we can do a relatively quick re-image," Velazquez said. "If a student wants to contest a grade, we have the ability to go to Panasas to find the database, reactivate and see exactly what the student's interactions with it were."

Panasas' underlying infrastructure is similar to the object storage platform Amazon used to build S3, but Panasas has also developed integrations with standard file protocol interfaces such as CIFS and integrated ActiveStor with Windows Active Directory, which Velazquez said made deployment of the system simpler. Cloud data storage systems that expose the object storage interface require application development so they can be integrated into the enterprise data storage environment.

While the infrastructure is performing well so far, Velazquez said there's room for improvement, particularly on Amazon's side. "If I make a change to a file, it can still take 25 to 30 minutes to get replicated across [Amazon's] CDN," he said. "I wish there was a way to at least indicate which systems might not be fully consistent in the cloud."


Dig Deeper on Hybrid cloud storage

Start the conversation

Send me notifications when other members comment.

Please create a username to comment.