Telco Systems announced that it has achieved a performance milestone in the development of a full Open vSwitch (OVS) data path hardware offload solution.
The company said its performance tests showed a three-to-five-times increase in the packet throughput, while also lowering the latency rate by double when compared to a software-accelerated OVS-Data Plane Development Kit (DPDK) data path with a fully deployed virtual network function (VNF.)
Additionally, Telco Systems said the new solution was a step forward in regards to removing major performance bottlenecks and freeing valuable CPU resources by fully offloading all host virtual switching to the programmable hardware engine integrated on the system on chip (SoC) itself, while providing integration with Open vSwitch control plane, OpenStack and the management and network orchestration (MANO) layers.
It's also able to free up one or more CPU cores that are typically consumed by host DPDK for use by VNFs, which allows more VNFs to be run or even higher VNF performance to be achieved without upgrading the hardware.
Telco Systems' new solution uses the NXP Semiconductors' NXP Advanced I/O Processor (AIOP) engine, and is integrated on NXP Layerscape LS2088A and LS1088A multicore processors.
RELATED: Telco Systems strikes a deal for uCPE with Arm
"With our groundbreaking OVS HW offload solution, customers will be able to achieve the performance required by large enterprises or edge compute use cases at the cost of a small-medium uCPE," said Telco Systems CEO Ariel Efrati, in a statement. "Significantly lowering the latency also enables our NFVTime solution to support next-generation 5G/MEC use-cases. This technology is strategic for the market as the requirement is driven by leading Tier-1 customers.
"Telco Systems continues working closely with NXP and its Arm-based SoCs to further optimize and extend its hardware acceleration solution, making it available to VNF vendors as a generic offload API."