New Systems to Strengthen Your Core


Virtualization, cloud computing, network consolidation and the relentless expansion of data loads in general are pushing enterprise networks to the limit. And while new technologies like I/O virtualization are helping to alleviate some of the pressure out on the edge, core systems are facing unprecedented burdens in the effort to get data where it needs to go.

So it's no surprise that we are seeing a flood of new core switching technologies aimed not only at boosting throughput but enabling new levels of scalability and density as well, all while keeping within available power budgets.

On the silicon level, Broadcom just rolled out the XGS Core fabric architecture that the company is making available to OEMs in the enterprise, data center and service provide channels. The device consists of a traffic manager/fabric interface module, a scalable switch fabric and a high-density buffer manager, relieving the need to combine individual ASICs, FPGAs and other components to produce a universal fabric architecture. The switch scales up to 1 Tb per line card with support for 100 GbE speeds. The company also provides a "virtual chassis reference systems" that provides software, schematics and documentation that allows OEMS to more easily integrate the XGS switch into their own system configurations.

3Com, meanwhile, is leveraging its H3C portfolio to go after the large networking needs of top-tier enterprises. The latest effort is the S12500, an aggregation/core switch system that can deliver 2.2 billion packets per second using a 6.6 Tbps non-blocking fabric consisting of either 128 10 GbE ports or 864 single GbE ports. The system also manages to get power consumption down to 80 watts per 10 GbE port, about half that of existing solutions. It also provides seamless transition between 40 and 100 GbE and FCoE traffic and is outfitted with the company's Intelligent Management Centre (IMC) single-pane management platform that is optimized for heterogeneous networks.

And despite its ongoing reorganization, Nortel is still looking to stay in the game with the Virtual Service Platform 9000, which the company has slated for release later this year. The system scales up to 27 Tbps in a single-chassis configuration, but can hit a whopping 100 Tbps in a quad architecture. Each chassis offers 10 slots for any combination of 24-port/10 GbE, 48-port/1 GbE or 48-port/10/100/1000 Mbps modules. At 10 GbE, system density is 240 ports per chassis or 720 per rack. It also supports Nortel's Split Multi-Link Trunking system that aggregates multiple physical links for simplified networking and improved load balancing.

If current trends continue, enterprise spending will focus more and more on the networking side of the house rather than the server/storage side. Virtualization and cloud computing take the onus off of capacity and resource availability and place it directly onto throughput and data flexibility to ensure that the enterprise can handle its growing workloads. Core switching, then, becomes the new ground zero in the drive to turn the data center into a competitive advantage.