EMC's Proven Solutions team continues to amaze me with the quality of work they are producing including building out some challenging scenarios that are meant to replicate a customer's environment. By doing this they can test the scenario, understand the expected results and document EMC's recommendations for properly deploying the configuration.
Last month the team produced a solution highlighting how to virtualize Microsoft Sharepoint Server 2007 using Hyper-V R2 and using EMC's RecoverPoint Cluster Enabler technology for high availability as well as local and remote data protection. Here's the breakdown...
The team used 6 physical servers and then created virtual machines (VMs) to host 6 web front end servers, excel services, index servers, application servers and domain controllers. Sharepoint Server is a great application to virtualize since there are so many different components and by virtualizing them you can easily consolidate them on physical hardware and then easily coordinate high availability through Microsoft's clustering technology. Overall 1.5TB of Sharepoint farm user data was created to test with.
The solutions team used a pair of CLARiiON CX4-240s with RecoverPoint appliances and the Cluster Enabler software. Cluster Enabler works with Microsoft clustering and allows you to stretch an existing Microsoft Failover cluster to provide high availability in separate geographical locations. In other words the nodes of the cluster physically reside in different sites. RecoverPoint is a replication and data production solution that provides continuous remote replication, continuous data protection and concurrent local and remote data production for SAN volumes. RecoverPoint is an out of band appliance that provides data reduction and compression so data is replicated across long distances while using less bandwidth.
A diagram of the overall solution is here:
So what was tested? The team conducted a series of tests including performance tests and tests before and after replication while simulating three key user activities of browse, search and modify. They also performed full farm failover of the entire environment from one site to the another to understand the steps involved and document the time to failover, bring the farm back online and then to failback.
And what did they find out?
First of all full site disaster recovery caused only 7 minutes for the connected users as seen in the following graph:
Other key findings included:
- Planned failback of the SharePoint had a cumulative downtime of only two minutes.
- Since RecoverPoint is an out of band appliance solution, the impact of using the RecoverPoint splitter on the CLARiiON array was ~1 percent under full Sharepoint full load. This means that the storage array can continue to do what it does best which is meet the I/O demands of the application while the dedicated appliances handle the replication process.
- Bandwidth was reduced 4.2 times due to the bandwidth reduction and compression technologies in RecoverPoint.
- By using Hyper-V the team was able to consolidate 15 physical servers down to 6 virtual servers.
In summary, Hyper-V performed great for this scenario and helped to reduce the number of servers in the environment. Using EMC's RecoverPoint and Cluster Enabler failover was performed in minutes with very little impact to the storage, servers or network. Using these technologies a customer can create a highly available virtual Sharepoint Server environment.
For more information be sure to check out the whitepaper and stay tuned for more exciting Proven Solutions using Hyper-V from our team of experts!
Comments