VMware previewed the upgraded Site Recovery Manager at VMWorld in August, but did not set a release date then. The application was informally known as Site Recovery Manager 1.5 until today but it is officially called version 4 to keep in line with VMware's vSphere virtualization platform. Site Recovery Manager 4 is priced the same as the initial version of Site Recovery Manager that launched in mid-2008 -- $1,750 per processor.
Although it still lacks automated failback and other features, Site Recovery Manager 4 can take advantage of vSphere 4 capabilities such as fault tolerance and lets customers replicate files through NFS as well as block storage. It also supports automated failover from multiple production sites into a single shared recovery site.
The new features for Site Recovery Manager come as organizations look to expand disaster recovery for their virtual server environments.
"Customers with vSphere 4 are looking at virtualizing their production applications more," VMware product marketing manager Jon Bock said. "Customers are starting to include DR as a checklist item, and we see SRM as a differentiator from what other virtualization providers have."
Major league baseball's Tampa Bay Rays have been using SRM since it first came out last year. The team backs up data from its minor league affiliates, baseball academies and other offices around the country, in Venezuela and the Dominican Republic. Those sites replicate to the Rays' main data center in Tampa and the team uses its spring training site in Port Charlotte, Fla., for its DR site, Rays senior director of information technology Juan Ramirez said.
That makes the shared recovery site addition especially valuable, Ramirez said.
"We have many smaller sites to protect, so the many-to-one-failover really helps," he said. "We're in a very sensitive area for hurricanes. We test every production group once a month. We bring over our SQL and Exchange stores, and we validate by failing over. Now I feel comfortable in our ability to recover."
The Rays use EMC Corp. storage, and Ramirez said coaches bring a 4 TB Iomega network-attached storage (NAS) box on the road to store videos of "every swing and every pitch." When the team returns, the Iomega data is synchronized to the main storage system and replicated to the DR site.
"We use Site Recovery Manager with EMC replication manager to take snapshots of all of our virtual machines," Ramirez said. "Before SRM we only replicated file systems to our DR facility. That was a lot of manual work, keeping up with all the policies and manually verifying files were readable and mountable. It was hard for us to maintain."
Ramirez said the NFS support also helps him. With Site Recovery Manager 1, he said he converted NFS files to iSCSI. "NFS support lets us access files and copy them over to a different host and take individual snapshots with Microsoft VSS [Volume Shadow Copy Service]," he said. "With iSCSI we can't see those files unless we mount them to the DR site. With NFS I can see them and manipulate them better."
VMware Site Recovery Manager still missing pieces
Burton Group analyst Chris Wolf said Site Recovery Manager is an area where VMware is far ahead of other virtual server providers but sees areas for improvement.
"The multisite piece is a huge step forward, even for service providers who want to offer DR as a service." he said. "Nobody else is even competing in this space. VMware is defining this space."
Still, Wolf said there is always room to "nitpick data protection products" and while SRM "does a lot of good, it can get better."
Wolf said automated failback is an important piece of disaster recovery, and he would also like to see VMware add intelligence for handling certain files left behind when recovering from a disaster. He said files such as page files and temp files aren't needed to recover virtual machines but must be replicated anyway because VMware looks for them when recovering.
"It needs to create placeholders for those files on the alternate site," he said. "Some VMware storage partners have started to talk about that issue."
Ramirez said, "Failback is a huge undertaking if we have a disaster. They have a plug-in [for EMC Celerra], but I believe it should be part of SRM." He would also like to see better reporting. "They've improved reporting, but there are always additional things we can provide to executives," he added. "Something that says, 'This is the process' and 'this is the end result.'"