It should be clear to most of us by now that server virtualization changes the disaster recovery game dramatically. Rather than having to maintain a server at your DR site for each server in your production environment, you can replicate physical, and/or virtual, servers from your production site to virtual servers at your DR site, reducing the cost of protecting production systems or increasing the number of servers you can protect.The fly in the ointment has been that the virtual server environment at your DR site would have to handle your production loads in the event of a disaster, so you couldn't stack too many virtual servers on a single host. SunGard's new business continuity services for VMware addresses this problem by using different virtual server hosts during normal operations and during a disaster.
Even though the solution relies on host based replication services during normal operations, they, and their disk subsystems, are pretty lightly loaded. Their applications, including database and messaging servers, are suspended and they only need to process replication traffic. Production servers also need to deal with database queries and disk reads. That means SunGard can pack more virtual servers on each virtual server host, even combining virtual servers from multiple customers on the same host, and can use lower cost high-capacity SATA disk arrays to support these servers.
When a customer declares an emergency, SunGard uses VMotion and Storage VMotion to move the customer's virtual machines, and their data, to a production environment that has enough compute horsepower and disk performance to support the user's production loads.
If SunGard's pricing model reflects the rationalization of costs that this architecture creates, this could be a great solution for the SME (small and medium enterprise) market.
SunGard plans to expand its offerings to include VMware's Site Recovery Manager, and array based replication, when that application ships in the next few weeks.