Method and apparatus for achieving fault tolerance in packet switching systems with inverse multiplexing

A packet switching system, comprising:

• a core switching network having kN input ports and kN output ports, the core switching network providing a plurality of switching paths for data packets between the kN input ports and the kN output ports, each of the plurality of switching paths connecting one of the kN input ports to one of the kN output ports;
• a group of N input controllers, each of the group of N input controllers being attached to k input ports of the core switching network and providing an interface for an input communication link carrying an input stream of data packets, each of the group of N input controllers being configured to partition the input stream of data packets into a sequence of asynchronous frames using a sequence of marked data packets prior to transmitting a resulting stream of data packets into the k input ports of the core switching network;
• and
• a group of N output controllers, each of the group of N output controllers being attached to k output ports of the core switching network and providing an interface for an output communication link carrying an output stream of data packets, each of the group of N output controllers being configured to reassemble a received stream of data packets from the k output ports into the output stream of data packets.