Project Spotlight

University of San Diego Implements Deduplication Appliance to Back Up Highly Virtualized Environment

• By Leila Meyer
• 01/08/15

The University of San Diego has virtualized nearly all of its IT infrastructure and has now dramatically reduced the backup times and storage requirements for its virtual environment, all while saving money.

About five years ago, USD began investigating virtualization to improve the efficiency of its processing, storage and data center technology. The IT team invested in VMware and built an infrastructure that enabled the university to be highly virtualized for its Windows and Linux-based systems. The team continued to research virtualization technology and refine its virtualization strategy, leading it to implement a Vblock System 300 series from VCE, which includes networking, storage technology and software.

The university's infrastructure is now 98 percent virtualized, and the only reason it isn't 100 percent is because the remaining portion of the university's IT infrastructure requires physical hardware or otherwise cannot be virtualized.

Challenge

While USD's highly virtualized environment improved its efficiency, it generated so much data that backups couldn't complete within their designated time windows. The university was backing up more than 200 terabytes on a weekly basis, and simply adding more hard drives to the backup system was no longer a cost-effective option. That was when Mike Somerville, manager of system services at USD, began researching deduplication systems to compress the volume of data being backed up, so the backups could complete on time.

In its search for the best fit, USD had two fundamental goals: "The first goal was to be able to complete our full and incremental data backups in a timely way," said Christopher Wessells, vice provost and CIO. "And secondly, find a cost-effective solution that would allow us to back up and restore our data without spending massive amounts of money on ever-increasing demands for more storage."

Somerville had more specific goals: "Ease of installation, ease of management and ease of use," he said. "I have a finite number of people who run our infrastructure, so I don't want to install cool technology for technology's sake and then have to train a whole bunch of people on something that is simply too difficult." While some of his highly technical staff members would be using the system, he also wanted other, less technical people to be able to log on and perform a rudimentary restore if necessary.

Solution

The team settled on a Quantum DXi6700 Series deduplication appliance with vmPRO software. "It was an ideal fit for an institution of our size and the total data needs of the institution," said Wessells.

The team at USD collaborated with Quantum and Partners Data Systems to implement the solution in less than a week. "That involves it physically being on site — allocating space and talking with facilities and getting power and that sort of thing," said Somerville. "Once the system was powered up, it was several hours after that that it became functional."

USD is a liberal arts institution with four centralized data centers, and the Quantum system manages the entire university's backups. The vmPRO software handles the backup and restoration of all of the Windows and Linux virtual machines, and the university uses CommVault software to manage the backup and restoration of data that is stored on network-attached storage.

The vmPRO and CommVault software control the flow of data to the DXi6700 deduplication appliance, which intelligently compresses the data by eliminating duplicate copies of any data. The system uses inline data deduplication, so the DXi compresses the data as it stores it. "So it's not like I need to take 20 terabytes of data and put 20 terabytes of data somewhere and then manipulate it after the fact," said Somerville. "I simply copy the 20 terabytes of data into this system, and it ends up being one terabyte of data."

The backups are stored on the DXi appliance itself, and once they reach a certain age, they are moved off the device to inexpensive tape backup for long-term offsite archival.

Results

The team at USD has been thrilled with the results of the implementation. The system compresses the backup data at a ratio of 20-to-1, which saves storage space. "Right now, the actual amount of space we're using is just about 10 terabytes, which means that we're backing up almost a quarter of a petabyte of actual data," said Somerville. And the system uses 256-bit encryption to ensure that the backed up data is secure.

The system has also dramatically reduced the amount of time required for backups. "It used to take many hours and now it takes about one hour or less," said Somerville.

Somerville and Wessells expect the system to scale with their data growth for years to come. They are currently using 10 terabytes of the system's 32 terabyte capacity, and it can expand to 80 terabytes. "That's future-proofing us for the next few years for sure," said Wessells.

By the time they max out their 80 terabytes, Somerville has "no doubt that the model that they will provide us will grow to much larger than 80 terabytes."