Virtual Networking Fuels Multi-Rate Switching

Virtual networking will likely deliver entirely new levels of flexibility for data operations. At the same time it will convert network management from a distinct engineering-oriented process to an element of application coding.

But regardless of how the actual resource provisioning is done, at some point data must traverse through wires, ports, routers and other real things in order to fulfill its user demands. And that means the enterprise still has a vested interest in making sure its physical layer infrastructure is as amenable as possible to the widely divergent forms of networking that modern data environments require.

One way to do that is through multi-rate network topologies. By accommodating various networking speeds on dedicated hardware, the enterprise can foster greater flexibility in its software-defined architectures while still keeping network infrastructure to a minimum.

According to Light Reading’s Simon Stanley, optimal core data center networking performance for cloud services and virtual networks requires low-latency switches with 25 Gbps serial interfaces and both 25 GbE and 100 GbE ports. When combined with software-defined networking (SDN) and network function virtualization (NFV), as well as third-party or open-source operating systems and in-service feature upgrades, organizations should gain the service-level flexibility to accommodate virtually any workload that traditional or emerging digital data environments can produce. Current chipsets are available in the 3-6 Tbps range of aggregate bandwidth and can be programmed in the field using new languages like P4.

On the processor level, Broadcom is out with a new dual-port PHY for 10, 25, 40, 50 and 100 GbE interfaces, which the company says is suitable for high-speed, mission-critical configurations for enterprise, government, cloud and service provider networks. The device supports a 25G serializer/deserializer core and the IEEE 802.1A media access controller security (MACsec) standard with 256-bit AES encryption. Other features include a low-power 28 nm CMOS design and long-reach performance capable of compensating for more than 35 dB of insertion loss at 25 Gbps.

On the edge, most organizations will have to segment its network connectivity below 25 GbE across both data center and campus networks, which is why companies like Extreme Networks are pushing flexible data rates on their latest devices. The company recently released several additions to the ExtremeSwitching portfolio, including a pair of multi-rate devices that support IEEE 802.3bz and NBASE-T Alliance Ethernet connectivity at 1, 2.5 and 5 GbE rates. The company says its full portfolio not only provides flexible network provisioning for dynamic workloads but a predictable migration path to higher port densities.

Meanwhile, Microsemi and Aquantia have devised a new reference platform to coordinate switch development on both the hardware and software levels. The design integrates Aquantia’s PHY API and firmware into Microsemi’s Linux-based application software solutions to support 24 2.5 GbE and up to four additional 2/2.5/5/10 G BASE-T port configurations. The system also provides low jitter and PoE capabilities that will become increasingly important as Ethernet infrastructure integrates more directly to wireless LAN and device-level WiFi solutions.

Network flexibility is about more than just finding the fastest route from point to point. Its aim is to produce an optimal networking environment in terms of connectivity and cost so that the enterprise gains the highest level of performance with the greatest resource efficiency.

Through capabilities like multi-rate switching, organizations can tailor throughput to the needs of the application without having to pay for underutilized bandwidth.

Arthur Cole is a freelance journalist with more than 25 years’ experience covering enterprise IT, telecommunications and other high-tech industries.

Latest Articles

Follow Us On Social Media

Explore More