Separation of control plane functions using virtual machines in network device

摘要

Techniques are described for separating control plane functions in a network device using virtual machines. The techniques include initializing multiple virtual machine instances in a control unit of a standalone router, and running different control processes for the router in each of the virtual machines. For example, in a root system domain (RSD)-protected system domain (PSD) system, a control unit of the standalone router may support a RSD virtual machine (VM) and one or more PSD VMs configured to form logical devices and execute logically separate control processes without requiring physically separate, hardware-independent routing engines to form the PSDs. Each of the RSD VM and PSD VMs includes a separate kernel, an operating system, and control processes for the logical device. When a software failure occurs in the PSD VM, the PSD VM may perform a software failover without affecting the operation of the RSD VM.

主权项

1. A method comprising:
initializing a first virtual machine in a control unit of a network device that operates as a logically separate routing engine assigned to one or more packet forwarding components of the network device to form a logical device, the first virtual machine including a first kernel with a first guest operating system, and a first set of control processes for the logical device; initializing a second virtual machine in the control unit of the network device that operates as a routing engine for the network device, the second virtual machine including a second kernel with a second guest operating system, and a second set of control processes for the network device that map assignments of the packet forwarding components to the first virtual machine; performing storage disk input/output operations with the second guest operating system included in the second virtual machine for data generated by the first set of control processes in the first virtual machine; and when a software failure occurs in the first virtual machine in the control unit, performing a failover of the first virtual machine without affecting the second virtual machine.