EMC's VPLEX enables data replication over long distances
Latest storage arrays can replicate information when more than 2,000 miles apart
By Lucas Mearian | Computerworld US | Published: 15:20, 11 May 2011
The company also unveiled an integration of its Cloud Tiering Appliance and its midrange SAN/NAS VNX array, which allows older or unused data to be moved off the array to a lower-level tier of storage.
The EMC Cloud Tiering Appliance supports archiving to secondary storage with file retention capability, such as Data Domain, and will provide automated policy-driven migration, which allows users with big data challenges to automate data migration from multi-vendor environments to the VNX or to EMC Isilon clustered NAS storage.
Related Articles on Techworld
Moving data was a big theme at this year's show, from placing it in virtualised storage and migrating it among tiers of storage to replicating it offsite for collaboration or disaster recovery.
At its user conference last year, EMC announced vPlex Metro, which allows synchronous replication of data over distances of up to 100 kilometers.
Today, the company announced VPLEX Geo, which allows data to be replicated between storage arrays over greater distances. Over the next 18 to 24 months it expects to announce VPLEX Global, which would allow data to be shared anywhere around the globe in real time, said Brian Gallagher, president of EMC's Symmetrix & Virtualisation Product Group.
EMC partnered with several companies, including Cisco, Brocade, Silver Peak and Ciena, to use their WAN optimization technologies to achieve the data replication over distance.
In a question and answer session with press and analysts, Gallagher said the box only replicates data changes or updates and can only withstand a 50-millisecond delay between servers in geographically dispersed data centres.
"When an application comes through the [storage area network], the write is committed to disk as well as replicated to the remote site," he said.
The VPLEX appliance's "secret sauce" is its ability to tell applications such as Exchange, Sharepoint, or Oracle, that they're receiving the data in real time.
Cancer Treatment Centers of America (CTCA) just went live with EMC's VPLEX Metro technology. The hospital system is using VPLEX for high availability between redundant SANs - an EMC VMAX Symmetrix and an NS960 array - in the same data centre rather than failing over to a secondary site - for now.
"We saw that if our VMAX was to fail, we needed some other type of redundancy," said Eric Sato, manager of server administration at CTCA. Sato said his company is considering using VPLEX Geo and Global eventually to replicate data to its brand new secondary data centre in Madison, Wisconsin.
CTCA hired consulting firm Ahead to help it with the VPLEX implementation. Eric Kaplan, president of technical consulting at Ahead, said CTCA is not atypical of many businesses that are looking for greater resiliency.
Limitations caused by the speed of light typically cause applications to time out when data is being replicated between machines over distance. This means that if a machine sending data does not receive confirmation from the recipient machine within a certain span of time, it considers the computer operation failed.
Typically, vendors have solved this problem by sending data asynchronously over distances, so that a response message is not required from the recipient machine. However, asynchronous data replication means that if a disaster occurs, causing a primary data centre to go down, the data in transit to the secondary data centre is lost. In order to share data in real-time, synchronous replication is required.
EMC's VPLEX appliance basically cheats speed of light limitations by sending across a light-weight code with the data to invalidate the requirement for a fast response between servers in multiple sites - something EMC calls "cache coherency".
The appliance rests between the storage network switches and the storage array.
Gallagher said the VPLEX appliances can scale from two to eight nodes or locations. However, when it comes to clustered applications, such as Oracle RAC or VMware HA replication distances are limited to 100 kilometers, he said.
In addition to the new functionality, EMC also reduced the size of the appliance by using smaller rack-mounted boxes with more powerful Intel multi-core processors, PCI Gen 2 cards for faster throughput, and new 10 Gigabit Ethernet WAN interfaces for faster communications between clusters.
EMC is charging $100,000 (£60,600) for an appliance with 1TB of capacity on it, but Gallagher said street prices would vary depending on the size of the implementation.
In related news, EMC also announced has added support for native Fibre Channel over Ethernet (FCoE) connectivity on its high-end Symmetrix VMAX array.
EMC competitor NetApp added FCoE capability to its high-end Fabric-Attached Storage (FAS) arrays last year through an upgrade in its Data ONTAP 8 operating system. Around the same time, EMC also added FCoE to its midrange line of VNX SAN/NAS arrays.
EMC said it was important to add the technology to address the trend toward the convergence of data networks in corporations today.
Additionally, EMC said it increased security on its Symmetrix VMAX with new support for external servers using RSA Data Protectio Manager, which encrypts data at rest and stores and manages the encryption keys.