JUNOS High Availability

Hardware Capacity

One of the first topics in any discussion of network scalability is hardware. It is imperative that hardware selected for the network can meet the throughput needs and withstand the processing burden of the network. In networks with high availability requirements, chassis are often deployed in pairs. We discuss cost and design implications of pairing devices in Chapter 1, and in this chapter we consider the 50% rule to be gospel. For a high availability network in which chassis are deployed in pairs and in which the load is balanced across the redundant pair of switches, routers, firewalls, and other devices, the 50% rule means that no more than half the resources or capacities of either device in the pair should ever be in use at the same time. If this threshold is exceeded, and especially if it is exceeded consistently, and if one device in the pair fails, the remaining device will be unable to assume the load of both devices and the redundancy scheme will fail.

Device Resources to Monitor

All JUNOS-based products provide functional separation between the control plane, which operates in the routing engine (RE), and the data plane, which operates in the Packet Forwarding Engine (PFE), as discussed in Chapter 3. Trivial Network Protocol (TNP) enables communications between the control and data planes. You must understand the resources and capacities of the RE and the PFE to properly select hardware for a scalable network and to monitor each chassis as the network ...