IBM has revamped its remote replication technology to make moving data over distances faster and less expensiv...
Big Blue announced Tuesday that its new software package, dubbed Global Mirror for Enterprise Storage, uses its asynchronous PPRC (peer-to-peer remote copy) to speed disk mirroring over distances of more than 300 kilometers. IBM claims it now outpaces competing technology from EMC and Hitachi Data Systems by lowering the lag time of data in transit to just five seconds.
EMC refutes IBM's claims of superiority. EMC spokesperson Dave Farmer says IBM is citing "outdated information posted on EMC's website, which is currently being updated." Farmer said EMC's SRDF Asynchronous (SRDF/A) software also replicates data in five seconds over unlimited distances. HDS could not be reached for comment by press time.
Here's the rub on IBM's claims. Global Mirror requires two Enterprise Storage Servers – also known as Sharks – at either end of the transfer. They must be model 750, 800 or 800 turbo running version two of IBM's FlashCopy and PPRC software.
While Global Mirror can't achieve these speeds and distances when replicating to a cheap JBOD array, Mike Kahn, a managing director for the Clipper Group says IBM's new ESS 750, the baby of the Shark family, is "will lower remote costs, if an enterprise can live within its maximum capacity of 4.6 TB."
Kahn adds that Global Mirror's ability to extend to eight remote sites offers redundancy and geographical distribution. "IBM is extending PPRC to offer faster asynchronous mirroring between two or more sites in an open systems environment. The key words are faster, asynchronous, and more sites," Kahn said.
Global Mirror's three-to-five second lag time from the primary system is a serious boost when compared to the 10 to 30 seconds achieved with older methods. "This shortened time window reduces the risk of data being lost in transit, if there is a failure," Kahn said.
Global Mirror replicates data asynchronously from one Shark to another at a remote site every three-to-five seconds. The remote Shark automatically makes a copy of the volume using IBM's FlashCopy feature for recovery purposes. IBM says just two links are required to perform the replication process. Fewer links means a smaller bill from the telecommunications company.
According to John Power, IBM's product strategy manager for Shark, Global Mirror saves the end user money because it makes more efficient use of available bandwidth and only requires two Fibre Channel links. "Most of the time competitive solutions require four or more Fibre links. We're seeing a 50% reduction in [telco costs] for an average configuration," he says.
Disaster recovery is typically done at local distances of up to about 300 kilometers. Beyond that, all vendors fall victim to the speed of light, causing lag time between when data is sent from a primary site and when it reaches the remote facility. Power said IBM obviously can't exceed the laws of physics, but can speed up the work that needs to be done on either end of the link -- hence the five second latency window.
IBM TotalStorage Global Mirror for ESS will be available for IBM eServer zSeries systems and open systems on May 28. IBM Business Continuity and Recovery Services will be incorporating Global Mirror as part of its rapid recovery portfolio.
New Sharks have the latest version of the software already installed at no extra price. However, if an end user is new to PPRC, there is a price for the software. Ballpark pricing for a brand new PPRC remote copy license of Global Mirror runs between $60,000 and $80,000. Power said the pricing model is tiered enabling IBM to adjust the bill based upon customer usage.