Cisco Delivers Redundant Arrays Of Independent Datacenters: The New RAID
February 23, 2010
If you walk into the gaseous mist of a private or public cloud, you will find multiple datacenters full of real hardware and software infrastructure. The huge clouds that span the earth would dissipate without all the servers to process the data, storage to hold the data and networking to connect the datacenters.
On February 8, Cisco announced the introduction of new products that are part of an industry-wide continuum of cloud infrastructure development. One Cisco product, Overlay Transport Virtualization (OTV), stands out by dramatically simplifying the deployment of Layer 2 connections between datacenters to form Redundant Arrays of Independent "Datacenters" (RAID).
RAID is an acronym commonly used to describe a Redundant Array of Independent Disks, a technology designed to achieve high levels of storage reliability and performance by dividing and replicating data among multiple disk drives. Data is distributed across multiple disks, but the array is seen by the computer user and operating system as one single disk. Similarly, with OTV, IT organizations can achieve higher levels of application reliability and performance by dividing and replicating data across multiple datacenters. Virtual machines and application workloads can now be distributed across multiple datacenters, but the array of datacenters is seen by the computer user and operating system as one logical entity.
Available in April on the Nexus 7000, the new Cisco technology is evolutionary but the potential effect on cloud development could be revolutionary. The concept of replicating data across datacenters is not new, but the process of connecting multiple datacenters is so complex and expensive that only business cases with the highest return-on-investment can justify the expertise and infrastructure needed to deploy MPLS or dark fiber.
What could drive a revolution in the development of Redundant Arrays of Independent Datacenters is the simplicity of deploying LAN extensions with OTV. According to Craig Huitema, director of marketing at Cisco Data Center Solutions, the product extends Layer 2 Ethernet LANs by encapsulating the traffic through the IP cloud. But what I find to be a breakthrough for IT is the set-up. The process of setting up OTV involves only five commands and five minutes of time. Compare that to weeks or months of architecture needed for an MPLS or dark fiber interconnect.
The first killer app for OTV, and dramatically simplified LAN extension, is VMware Vmotion. Datacenter managers love Vmotion because it allows them to migrate any virtual machine running any operating system across any type of server and storage supported by VMware ESX, including Fibre Channel SAN, NAS and iSCSI SAN. The only catch is Vmotion moves virtual machines only over a layer 2 network.
With OTV, large enterprises now have a choice of continuing to use MPLS and dark fiber or begin using OTV for new connections between datacenters. Additionally, representing another important breakthrough, for the first time small and medium enterprises have access to technology that makes it feasible to begin construction of a private cloud starting with applications like Vmotion that run across layer 2 networks spanning multiple datacenters.
The future of cloud computing requires Redundant Arrays of Independent Datacenters that deploy and operate as simply and reliably as Redundant Arrays of Independent Disks do today. OTV from Cisco represents an important step in the industry continuum of making that happen.