Infineta Systems’ WAN Optimizer Tackles High-Capacity Data Center Interconnects

Infineta Data Mobility Switch accelerates critical network traffic, reduces bandwidth consumption.

• 06/06/2011

Infineta Systems, a hyper-scale WAN optimization systems provider, has released its Infineta Data Mobility Switch (DMS). The DMS provides 5- to 10-fold bandwidth capacity gains, IP traffic acceleration, and critical application prioritization and assurance over multi-gigabit WAN links (the hyper-scale WAN). The DMS operates at 10Gbps wire speeds, supporting the next generation of advanced data center architectures that leverage key innovations around agile, high-performance computing, server, and storage services delivery models.

Infineta DMS combines fine-grained quality of service (QoS) controls, advanced network-layer acceleration technologies, and the Velocity Dedupe Engine (VDE) -- a hardware-based, data-deduplication (“dedupe”) system core -- to deliver maximum throughput for high-capacity WANs. With the DMS, for example, customers can turn a 1Gbps WAN into a virtual 5Gbps link. Because the DMS can be configured to guarantee minimum data rates for individual classes of service, customers can carve out “protected circuits” for their bandwidth-intensive, latency-sensitive applications over their existing WAN links. Similarly, an OC12 (622Mbps) circuit can be turned into a 3Gbps link and an OC48 (2.54Gbps) circuit into a 10Gbps link.

Customers enjoy increased network throughput and guaranteed bandwidth availability. Enormous traffic growth can be supported without requiring WAN upgrades, saving both CAPEX (WAN equipment upgrades) and OPEX (incremental bandwidth costs), with typical return on investment in less than a year.

Over the last decade, the combination of IT consolidation and end-user dispersion has led to greater distance separating users from application servers and has become a major pain point. User dissatisfaction and plummeting productivity were further exacerbated by the underlying protocols used by a majority of enterprise applications that were never intended to run over limited-bandwidth, high-latency WANs. Many software-based WAN optimization appliances have tried to address this application responsiveness problem faced by branch users.

Branch WAN optimization tackles what is described in networking jargon as the “north/south” traffic pattern -- the pattern of network traffic as it enters a data center from a remote location and makes its way to a server resource, then back out. The focus is on optimizing end-user connections, often numbering in the thousands or tens of thousands, across low-bandwidth, high-latency WAN links. Although the nature of the branch WAN optimization problem continues to evolve the core challenge remains the same: optimize connections between users and a diverse set of business applications hosted in a remote data center.

“East/West” Traffic and Hyper-scale WANs

Driven by the rapid adoption of server virtualization, the advent of the “cloud,” and business continuity technologies, today’s data center is undergoing a fundamental transformation. This shift is readily apparent on the data center LAN, where network architects are abandoning multi-tier, switching architectures in favor of a “flatter” network to better support the high-volume, low-latency communication prevalent between resources. These new traffic flows are best described as “east/west” in nature, with connections running between servers, storage arrays, and other network devices. Compared to “north/south” traffic, “east/west” traffic is distinct, limited only by the capacity of ever-scaling “machines” such as computing and storage systems.

Additionally, “east/west” traffic is no longer confined to a solitary data center. “East/west” traffic growth has led to it spilling onto the WAN connecting multiple data centers. Enterprises are realizing that treating their data centers as a unified resource pool provides greater availability, flexibility and cost-efficiency. In this interconnected data center model, the Hyper-scale WAN emerges as the lifeline that seamlessly connects resource pools across data centers.

Key requirements for the hyper-scale WAN are unique and straightforward. Improving inter-data center WAN performance invariably entails delivering improved scalability for critical applications (such as replication, long-distance “live” migrations, massive file transfers, and “Big Data”) that require high-capacity, expensive WAN circuits that must increasingly support more and more traffic volumes.

Infineta’s Data Mobility Switch

Infineta’s DMS has been purpose-built for the hyper-scale WAN and delivers the following features:

• Wire Speed, High-Efficacy Data Reduction: The DMS reduces multi-gigabit traffic by as much as 5x or more, while introducing a port-to-port latency overhead of only a few 10s of microseconds
  
• Extremely High Per-Flow Throughput at Any Distance: For throughput-intensive, bursts of traffic, the DMS supports 1Gbps+ speeds per flow at very large network latencies, making flow throughput independent of WAN latency
  

• Packet Loss/Congestion Avoidance: By managing WAN resource allocation at the transport level instead of relying on archaic packet-drop-mitigation techniques to police send rates, the DMS prevents congestion and contention due to multiple aggressive applications sharing the WAN
  

• Future-proof Scalability: With all DMS platforms able to support 10Gbps inputs, customers can choose to purchase a 2Gbps or 5Gbps model today and upgrade to 5Gbps or 10Gbps, respectively, at a future date through a software upgrade instead of requiring a disruptive forklift upgrade

DMS Pricing and Availability

The Infineta DMS is available in capacity increments of 2Gbps, 5Gbps, and 10Gbps; pricing starts at $80,000. All enterprises and service providers with two or more data centers can use DMS systems to instantly improve their high traffic network, while realizing a sub-year investment payback. For a full data sheet containing a specification chart and feature details, visit www.infineta.com.