Cisco's New FabricPath Reduces Large Networks' Need for Switches

Datamation content and product recommendations are editorially independent. We may make money when you click on links to our partners. Learn More.

As enterprise and datacenter networks become increasingly large and complex, vendors and networking standards bodies like the Internet Engineering Task Force (IETF) are aiming to make management easier. That’s where Cisco’s FabricPath technology comes in as a sweeping new approach to network infrastructure.

FabricPath is Cisco’s version of the IETF’s Transparent Interconnection of Lots of Links (TRILL) network management standard. Although TRILL remains still in development in the IETF, Cisco says FabricPath is capable of reducing the number of switches that are required for large server deployments while maximizing the available bandwidth.

Cisco spokesperson Shashi Kiran explained to InternetNews.com that protocols like spanning tree have restricted Ethernet domains’ ability to scale. The idea behind FabricLink is to get around those limitations: For example, Kiran said that on a large 2,048-server setup using spanning tree, 64 switches would have been required to route the traffic. With FabricPath, that same 2,048-server setup can be managed with only eight switches.

Kiran added that FabricPath is an extra feature coming in August for customers of the Cisco Nexus 7000 — Cisco’s core data center switch — in connection with the 5.1 release of Cisco’s NX-OS operating system.

Since the FabricPath technology is based on the as-yet-unfinished TRILL standard, Kiran said that Cisco will upgrade FabricPath to meet the standard once it’s complete.

Cisco isn’t the only vendor looking to include TRILL in its networking gear. Rival Brocade earlier this month announced its own networking effort that includes TRILL as a way to help scale networks.

Kiran noted that Cisco’s FabricPath approach will be able to run on heterogeneous networks with both Cisco and non-Cisco gear.

Cisco is also trying to package FabricPath technology with hardware and management in what it calls the FabricPath Switching System (FSS), an arrangement it positions as a way to provide massive networking scale. Kiran said that the top-end of the FSS can deliver up to 160 terabits per second of switching capacity, managing over 8,000 10-gigabit Ethernet switch ports.

Kiran added that the FabricLink technology is complementary to the Overlay Transport Virtualization (OTV) technology that Cisco introduced at the beginning of the year. OTV is intended to enable easier virtual machine mobility across data center networks.

In addition to Cisco’s efforts at easing network management, the company is also trying to make applications move faster too. Cisco this week announced improvement to its WAAS (Wide Area Application Services) WAN acceleration technology. WAAS is now available for the Cisco’s ISR G2 routers and is being improved for accelerating cloud-based application deployments.

However, WAAS is not directly available for Cisco’s Unified Computing Server (UCS). Instead, Kiran explained that a WAAS appliance can be deployed in front of a UCS box to help accelerate the traffic. He added that the market should “stay tuned” for future announcements about specific WAAS capabilities for UCS.

Hosted Cisco collaboration services for partners

Still, Cisco has at least one new UCS development to crow about: The company today also announced new hosted offerings for its collaboration services, powered in part by UCS.

“Through the Cisco Hosted Collaboration Solution, Cisco is enabling our partners to provide cloud-based solutions that offer unmatched levels of flexibility for their customers,” Barry O’Sullivan, senior vice president for Cisco’s Voice Technology Group, said in a statement.

Sean Michael Kerner is a senior editor at InternetNews.com, the news service of Internet.com, the network for technology professionals.