SafeAuto Steers Toward a Better Backup Strategy
By Samuel Greengard
For many companies, operating multiple data centers is both a blessing and a curse. The ability to replicate data and ensure that it's backed up effectively can prove critical—especially when a disaster or disruption takes place. However, ensuring maximum performance while avoiding problems such as latency and data synchronization issues can push an IT department to the edge … or beyond.
It's a challenge that Matthew Coy, senior director of IT infrastructure at SafeAuto knows all too well. In 2010, Coy identified a significant business problem: Two data centers, both located in Columbus, Ohio, represented a risk.
"If a wide-scale disaster occurred or a power failure took place, we had no way to keep the business operating effectively," he points out. As a result, the low-cost auto insurance provider, which offers state minimum coverage to drivers in 16 states, opted to construct a second data center located a few hundred miles away in Somerset, Ky.
SafeAuto opened the secondary facility in 2011. But what should have been a solution soon became a problem: "We had a serious issue with latency and how it affected network performance, particularly during database replication," Coy explains.
In some cases, an IBM DB2 database using high- availability disaster recovery (HADR) displayed errors that had to be repaired. The latency also caused problems with VMware replication and distributed file systems (DFS). What's more, in some cases, performance lagged within the claims center and on a number of key applications. In addition, SafeAuto wound up losing backup data between February and July of 2011 because of a problem with a telecom carrier.
After reviewing a number of vendors and solutions, Coy turned to Cisco Systems and implemented its Wide Area Application Service (WAAS) solution for WAN optimization. He found the price point agreeable, and the solution also provided a quick path to installation and relatively straightforward training requirements for the firm's IT department.
The decision wasn't easy because Cisco's WAAS solution was relatively new to the marketplace at the time. "We took a bit of a risk because other vendors offered products that had been in the marketplace a lot longer," Coy says.
However, he believes the approach made sense because the company had an existing relationship with Cisco and had already deployed the vendor's networking gear. Nevertheless, SafeAuto had to confront a number of technical challenges related to making the product work on its network.
In the end, SafeAuto was able to shift its network into high gear. The WAAS solution eliminated latency that had previously extended to about 9 milliseconds.
"We now have effectively zero latency," Coy reports. "There is no discernable lag in applications or database performance. The IT infrastructure behaves as it did when the data centers were only about 8 miles apart."
What's more, the company has experienced no disruptions or data loss. "We are operating on a far more efficient level than at any time in the past," Coy says, "and we have a network infrastructure that fully supports the business."