Research And Implementation Of IRF Stack System Based On Advanced Router

With the rapid development of network technology, increasing users and so extensive business cause an increasing number of network devices, and network expansion and network management even more difficult, then large-scale data center puts forward higher requirements for network forwarding capacity and reliability. The traditional solution is difficult to meet the needs of today’s high performance switched network, and can not solve the problem faced by fundamentally.For the problem of numerous physical devices, difficult management, and bad reliable stability in the network environment, this paper designs and achieves virtual technology based on distributed devices, namely the IRF Stacking technology. By stacking a plurality of distributed devices, you can assemble hardware resources and software processing capabilities of multiple devices to achieve collaborative work, unified management and ongoing maintenance of multiple devices. This article detailed designs the solutions of the core module of IRF stacking technology, the introduction of multi-stacked boards and interfaces achieves cross-board stacking, effectively prevents a paralysis of the entire system caused by fault of one physical interface; for cross block flows, load balancing enhances the overall system forwarding performance; using inter-board communication mechanism such as IPC and IBD process protocol packets and data packets between boards and frames, ensuring the correct operation of the top business between the IRF frames; by the technology MAD detection, it improves system reliability, and make the system in the event of a split to maintain maximum uninterrupted business operation; finally it’s detailed tested on a dedicated testing platform for system features and performance.According to the detailed function and performance tests of system, the test validates the system’s correctness of each module designed, satisfies the current demand for large-scale network communications, it has greatly improved for solving network currently faced by cumbersome network management, poor convergence performance, and low reliable stability.