Application non disruptive task migration in a network edge switch
AssigneeHome Run Patents LLC
InventorSajit Bhaskaran
Life of this patent
Estimating the term…
Abstract
An apparatus and a method are described for zero packet loss task migration in an network switch in a computer network. The invention relates to active or programmable networks, i.e. networks based on packet switching algorithms and switch configurations which are subject to change. A well-defined protocol enables an intelligent switch to migrate tasks from one forwarding engine to another without any packet loss. This enables the algorithms and configurations of the switch to be updated or modified arbitrarily.
Claims31
1-28. (canceled)
A network device, the network device comprising:
- a plurality of forwarding engines and a control processor, wherein the plurality of forwarding engines include a first forwarding engine and a second forwarding engine;
- a control processor, wherein the control processor migrates packet traffic to be routed to a destination by the first forwarding engine to the second forwarding engine;
- and
- a first switch connected to the first and the second forwarding engines, wherein during migration of the packet traffic from the first forwarding engine to the second forwarding engine, the packets are routed from the first forwarding engine to the second forwarding engine through the first switch with no datagram reordering.
A method of routing packet traffic via a network device, wherein the network device comprises a plurality of forwarding engines and a control processor, wherein the plurality of forwarding engines include a first forwarding engine and a second forwarding engine, and wherein a first switch is connected to the first forwarding engine and the second forwarding engine, the method comprising:
- migrating packet traffic to be routed to a destination by the first forwarding engine to the second forwarding engine using the control processor, wherein during migration of the packet traffic from the first forwarding engine to the second forwarding engine, the packets are routed from the first forwarding engine to the second forwarding engine through the first switch with no datagram reordering;
- and
- routing packets of the packet traffic to the destination using the second forwarding engine.
Dependencies are inferred from the claim text.