Juniper BNG CUPS High Availability
There are two aspects to Juniper BNG CUPS high availability, high availability for the BNG CUPS controller and high availability for the BNG User Planes.
BNG CUPS Controller High Availability
The BNG CUPS Controller consists of two micro services which run as pods on a Kubernetes cluster. The State Cache pod backs up all sessions and the BNG CUPS Controller state in high-availability mode.
In the case of a BNG CUPS Controller container failure, Kubernetes creates a new BNG CUPS Controller container. The BNG CUPS Controller gets its information from the State Cache container and builds a new state. After creating all the states, the BNG CUPS Controller reconnects to the BNG User Planes and continues from where it left off. BNG User Planes continue to forward traffic during a BNG CUPS Controller failure. No new logins are allowed until the BNG CUPS Controller recovers.
Figure 1 shows the BNG CUPS Controller container and the State Cache container.
BNG User Planes High Availability
High availability between BNG User Plane's routing Engines, also known as Graceful Routing Engine Switchover (GRES) is used in conjunction with BNG CUPS subscriber resiliency. On GRES, State and other information is replicated in a high availability mode across the routing engines. During GRES, the standby routing engine takes over as the active routing engine immediately.
For more information regard BNG User Plane high availability, see Use Juniper BNG CUPS Subscriber Groups.
Figure 2 shows the BNG User Plane high availability andwith GRES support between RE0 and RE1.
