Winckworth Sherwood Raises the Bar on Disaster Recovery Performance
Winckworth Sherwood’s data centre is located in its London headquarters, with a secondary disaster recovery (DR) site located 41 miles away in Bracknell. Daily, weekly, and monthly full backups and in-house SAN to SAN replication were done, with monthly tapes sent externally. The plan was to eventually replicate off-site but Christel Aguila, the Information Technology (IT) Manager at Winckworth Sherwood, had been looking for the right off-site replication technologies to meet the firm’s business continuity objectives for at least several years. Key concerns with the tape-based approach included both data loss on recovery (recovery point objective or RPO) and recovery time (recovery time objective or RTO). While Winckworth Sherwood was primarily a Windows shop using SAN-based storage from EMC, they did have some Linux servers and were selectively using both direct attached (DAS) and network attached (NAS) storage as well.
Ashvin Patel, the Infrastructure Manager responsible for selecting and deploying the DR solution, had invested in server virtualization technology from VMware, and was planning on using it to faciliate DR solution deployment. Patel’s plan was to replicate physical servers running key applications in the London data centre to Bracknell where virtual servers would be used to host critical applications – Microsoft SQL and Exchange – in the event of a disaster. “When using tape for DR purposes, recovering servers at the remote site was a very manual process that we knew would take at least several days,” said Patel. “In looking to use replication technologies, we expected that recovery would get much faster and easier because we would be working directly from disk, and we would be able to meet much more stringent RPOs.” Initially he evaluated SAN-to-SAN replication technologies, but was somewhat concerned about array vendor lock-in and bandwidth consumption.
When Patel first became aware of InMage, several features of the solution were attractive. “InMage’s ability to support heterogeneous storage gave us flexibility that SAN-based replication could not, plus they had integrated technologies like WAN optimization that provided for very efficient use of network bandwidth,” explains Patel. “And InMage’s architecture supported a local, disk-based recovery option too, giving us the ability to effectively use them to address RPO and RTO issues not only remotely but locally as well.”
Working with Nexstor, a local storage solutions provider, InMage did a demo for Winckworth Sherwood. They liked what they saw and brought InMage in to play around with it in their lab. “InMage demo’d impressively, but you don’t understand its full benefits until you see it in production,” adds Patel. Ultimately, Winckworth Sherwood ended up using InMage to address both remote and local data recovery, as well as application failover/failback (as part of the DR process).
“Using tape, recovering individual files locally generally required at least several hours, and in many cases significantly longer,” continues Patel. “Using InMage for local data recovery let us cut RTO in most cases to less than 30 minutes, and we now go to InMage first when we have to recover data.”
Having application-consistent recovery options was another feature on Patel’s wish list. Winckworth Sherwood had already been supplementing their daily tape backups with two VSS (Windows Volume Shadowcopy Services) snapshots to disk per day, but was excited that InMage had integrated their continuous data protection (CDP) technology with VSS right out of the box. “InMage gave us a very easy, fully automated, and low overhead way to generate more application-consistent recovery points for both SQL and Exchange, and we have now moved to an approach that uses VSS to create hourly recovery points,” says Patel. “Because of InMage’s integration of CDP and asynchronous replication technologies, we have access to these recovery points both locally and remotely, providing us with rapid, reliable recovery while at the same time significantly lowering our RPOs.”
Patel also noted the value that the InMage Analyzer, an I/O profiler included with the solution at no extra charge, provided in forecasting not only the recovery performance Winckworth Sherwood could support with their current network bandwidth allocations, but also in accurately predicting deployment costs. Now that the solution is in production, Patel continues to use the Analyzer to track network bandwidth, storage capacity utilization, and instantaneous RPO performance on an ongoing basis.
“The biggest benefit was being able to address our DR requirements with a flexible solution that significantly lowered our RPOs and RTOs, but its value for local data recovery and DR testing were important as well,” continues Patel. “Ultimately, it provided us with an insurance policy that would reduce downtime, keep our business running, and provide the peace of mind to know that our information assets are protected.”
InMage’s usability, particularly in restoring lost files, has proved very valuable. While the infrastructure team that monitors and manages the DR systems regularly find it easy to use, anyone shown key functionality can quickly restore files in minor emergency situations. “The speed at which we can provide partners and staff access to lost information is tremendous in comparison with traditional backup and restore systems,” concludes Patel. “We had been searching for the right cost-effective product for years to meet our recovery and business continuity objectives, and with InMage we now have the solution we were looking for in place.”