Telecom Platform Deployment Featured Article
High-Performance Telecom Platform Deployments Deliver Processing Throughput with Load Balancing
By Ashok Bindra, TMCnet Contributor
With the recent explosion in network bandwidth, applications like IPTV (News - Alert), traffic management and shaping, and other packet processing-intensive functions are demanding systems with high throughputs in order to avoid unnecessary latency in the network and deliver the quality of service that subscribers expect. With large amount of data per subscriber, these applications are demanding high bandwidth advancedTCA (ATCA) systems that have the ability to provide packet processing throughput and switch-based load balancing.
Fortunately, the state-of-the-art in ATCA system has evolved to the point where advanced switch, packet processing, and CPU functionality is becoming available and being integrated into leading-edge systems by companies like Radisys, a partner of telecom platform deployment provider NEI (News - Alert).
In fact, the performance of today’s high bandwidth ATCA systems have evolved from 10 Gbps to 80 Gbps. The result is faster time to market, lower cost per subscriber, and a stronger foundation for the new network applications of the future.
For instance, a recent white paper entitled, “High-Performance ATCA: Architectures for 80 Gbps/Shelf” written by Mike Coward, Radisys’ VP of Strategy & Innovation, shows that the deep packet inspection (DPI) ATCA system from Radisys is capable of handling 80 Gbps. When inserted into a network, it can perform inline inspection of a 40 Gigabit Ethernet network with full-duplex capacity, explains the white paper.
This system utilizes the eight 10G uplinks on the ATCA-FM40 base/fabric switch to connect into the network, and employs the load balancing features to distribute the incoming 80 Gbps across nine ATCA-PP50 packet processor blades running in an N+1 configuration.
Each PP50 blade is capable of processing 10 Gbps and performing flow matching. The blades send new flows to the ATCA-XE50 CPU blades for policy analysis. Once an XE50 has identified the flow, it communicates the policy decision to the PP50, and when subsequent packets for that same flow arrive, the PP50 can take action on the flow without involving the XE50 policy agents.
In addition, the PP50 can be programmed to forward the packet, drop the packet, reprioritize, modify the packet, calculate statistics, or duplicate the traffic and send it to an external monitoring agent for lawful intercept applications.
Edited by Jamie Epstein