Force10 Networks provides the line-rate, non-blocking Gigabit and 10 Gigabit Ethernet switch/routers that are integral to the infrastructure of scores of critical high performance computing systems. Force10 switch/routers are utilized in 6 of the top 15 and over 20 of the top 100 fastest supercomputers in the world.
Forschungszentrum Jülich, the New York Center for Computational Science (NYCCS) in Stony Brook, NY, and the Computational Center for Nanotechnology Innovation at Rensselaer Polytechnic Institute, all rely on the Force10 E-Series uncompromising resiliency, density and line-rate performance for their impressive IBM BlueGene supercomputers. With computing power of over 167 Teraflops (trillion floating point operations per second), the supercomputer JUGENE, at Jülich, has the brawn to take advantage of the E-Series vast switching capacity.
The fastest Dell cluster computers — at the National Center for Supercomputing Applications, NNSA/Sandia National Laboratories, Texas Advanced Computing Center and the Louisiana Optical Networking Initiative — utilize Force10 for their indispensible 1 & 10 Gigabit Ethernet networks.

Switch/Router key buying factors:
- Fault tolerant design enabling high application uptime even during Hitless component failover
- Deep buffering and congestion management technology ensures maximum network efficiency
- Ultra high density combined with system resiliency to ensure less network elements lowering capital and operating costs
[+|-]Where the E-Series Fits
Cluster Connectivity Requirement
Benefits of E-Series & High-performance Ethernet
User I/O
- High-speed (Gbps / 10 Gbps) & reliable connection to any router, firewall, or switch
- Predictable Layer-2 switching & Layer-3 routing in one device, simplifying network core and WAN topologies
- Flexible, cost-effective, standards-based Ethernet leverages existing knowledge base
Storage I/O
- No distance limitation for extending reach over the WAN
- High-speed (Gbps / 10 Gbps), reliable & flexible access to multiple storage sites
- Transport of SCSI (iSCSI) or Fibre Channel (FCIP) for Storage Area Networks (SAN)
- IP is the standard protocol for file level Network Attached Storage (NAS)
Management I/O
- Massive aggregation to capture existing Ethernet connections included on every server
- In-band management can overlay existing networks for high ROI, simplified topology
- Out-of-band management for maximum resiliency and security
Interconnect I/O
- High performance Ethernet is suited for all but the most latency tuned cluster applications (< 10 microseconds)
- Massive aggregation capacity per switch/router, >1,200 ports per chassis, more CPU connections per rack inch
- Simple, familiar and cost effective cabling
When the topology is all Ethernet
- High availability through zero packet loss hitless failover E-Series architecture
- Minimize cluster downtime through standards-based Layer-2 and Layer-3 resiliency protocols (802.1D/s/w, 802.3ad, ECMP)
- Simplify through density, all four I/O functions built around E-Series core, maximum scalability per chassis, longer product life cycles
- Lowest TCO through leveraging one technology, simplifying management, troubleshooting, sparing & training
back to top >