Would you entrust your mainframe to the cloud? Perhaps not for production, but IBM is hoping to gain customers for its cloud-based disaster recovery services by offering support for virtual mainframes. Currently, IBM offers cloud-based backup and disaster recovery services for the AIX, Windows and Linux platforms.
"We’re moving away from just backup to a replication environment in the cloud for all critical servers" -- including the mainframe, said Rich Cocchiara, distinguished engineer and chief technology officer for Business Continuity Recovery Services, during a recent one-on-one meeting at Computerworld's offices. "What cloud is doing is bringing the price down," perhaps to the point where more organizations may be willing to give up building or owning their own backup data centers. Instead of paying the capital expense of creating a backup data center, IT pays for access to virtual machines, as well as for the data backups and storage of the information.
The idea is to orchestrate end-to-end backups such that an entire data center -- both front-end and back-end systems -- can fail over quickly and reliably into a cloud-based backup data center. As systems spin up, connections between them are reestablished and everything, hopefully, comes back online without a hiccup. "The toughest part is the synching of the data and the referential integrity you have to have," Cocchiara said.
There are two key differences between IBM's cloud-based disaster recovery approach and its traditional service offerings: The virtualized backup resources can be either online or spun up very quickly, and customers can provision them on demand, without making a call to IBM.
But you can't do that with IBM's mainframe disaster recovery services just yet. Today, the client must call and request that the backup mainframe be brought online, and a IBM staff person must create a new mainframe partition for the customer. Cocchiara said IBM is working on mechanisms to let the user initiate the failover process without calling, and for IBM to automate that provisioning process on the back end. "In the near future you’ll probably see that,” he said.
When it comes to their mainframes, enterprise IT tends to be fairly conservative. Will it trust the cloud with its mainframe operations? Maybe, eventually. But many organizations also get value from their physical backup data centers by using those systems for noncritical functions such as application development when they are not in use for failover or disaster recovery testing. That's fine, said Cocchiara, as long as administrators resist the temptation to put production applications in the backup data center. "If you do that you end up with two production data centers and no real backup," he said.
IBM is also working on using analytics to anticipate outages and automatically fail over systems when the probability of imminent failure is high. "We want to say, 'You’re going to have an event. Let’s start the failover process now and if it occurs you’ll be ready.'" This application of analytics to risk, Cocchiara said, "is like a tornado early warning system that gets the business into a safe location before the event hits."