Managing processing utilization in a network node

A system for managing the utilization of processing resources in a packet-based network node comprising:

• a central processing unit (CPU) configured to learn forwarding information that is used to forward packets;
• a hardware-based forwarding engine configured to determine whether a packet should be sent to the CPU for learning;
• and
• a learning filter configured to receive packets from the hardware-based forwarding engine that are determined by the hardware-based forwarding engine to need learning and to decide whether to allow the received packets to reach the CPU based on whether any other related packets have already been allowed to reach the CPU;
• wherein related packets are packets from the same flow of packets and wherein the learning filter comprises a hasher that is comfigured to identify a flow to which the packet belongs and a state table for indicating whether a packet from an identified flow has already been sent to the CPU, wherein the hasher generates a hash value that identifies a flow and wherein the hash value is used to index the state table.

A method for managing the utilization of processing resources in a packet-based network node comprising:

• receiving a packet;
• sending the packet to a central processing unit (CPU) for learning;
• before the packet reaches the CPU;
• determining if a related packet has already been allowed to reach the CPU for learning;
• deciding whether to allow the packet to reach the CPU based on whether a related packet has already been allowed to reach the CPU;
• wherein determining if a related packet has already been allowed to reach the CPU for learning comprises:
• identifying a flow to which the packet is associated;
• determining whether a packet from the same flow has been allowed to reach the CPU;
• and
• allowing the received packet to reach the CPU for learning only if another packet from the same flow has not already been allowed to reach the CPU for learning.

A method for managing the utilization of processing resources in a packet-based network node comprising:

• receiving a packet;
• determing if forwarding information related to the packet needs to be learned to forward the packet;
• if learning is needed, deciding whether to subject the packet to learning based on whether any other related packets have already been subjected to learning;
• wherein deciding whether to subject the packet to learning comprises:
• identifying a flow to which the packet belongs;
• determining whether a packet from the same flow has already been subjected to learning;
• and
• subjecting the packet to learning only if it is determined that a packet from the same flow has not already been subjected to learning;
• wherein deciding whether to subject the packet to learning comprises identifying a flow to which the packet belongs and determining whether a packet from the same flow has already been subjuected to learning;
• and
• wherein identifying a flow to which the packet belongs involves hashing header information of the packet to produce a hash value and wherein determining whether a packet from the same flow has already been subjected to learning comprises indexing a state table using the hash value to obtain state information.

A system for managing the utilization of processing resources in a packet-based network node comprising:

• a central processing unit (CPU) configured to learn forwarding informatiion that is used to forward packets;
• a hardware-based forwarding engine configured to determine whether a packet should be sent to the CPU for learning;
• and
• a learning filter configured to receive packets from the hardware-based forwarding engine that are determined by the hardware-based forwarding engine to need learning and to decide whether to allow the received packets to reach the CPU based on whether any other related packets have already been allowed to reach the CPU;
• wherein the CPU and learning filter are located on a control module linecard and the hardware-based forwarding engine is located on a port interface linecard, the system further including a plurality of port interface linecards each having a hardware-based forwarding engine, wherein the learning filter is configured to receive packets from each of the interface linecards.

A system for managing the utilization of processing resources in a packet-based network node comprising:

• a central processing unit (CPU) configured to learn forwarding information that is used to forward packets;
• a hardware-based forwarding engine configured to determine whether a packet should be sent to the CPU for learning;
• and
• a learning filter configured to received packets from the hardware-based forwarding engine that are detemined by the hardware-based forwarding engine to need learning and to decide whether to allow the received packets to reach the CPU based on whether any other related packets have already been allowed to reach the CPU;
• wherein the CPU is located on a control module linecard and the learning filter is located along with the hardware-based forwarding engine on a port interface linecard, the network node furhter comprising a plurality of port interface linecards, each port interface linecard including a learning filter.

A method for managing the utilization of processing resources in a packet-based network node comprising:

• receiving a packet;
• sending the packet to central processing unit (CPU) for learning;
• before the packet reaches the CPU;
• determining if a related packet has already been allowed to reach the CPU for learning;
• deciding whether to allow the packet to reach the CPU based on whether a related packet has already been allowed to reach the CPU;
• wherein determing if a related packet has already been allowed to reach the CPU for learning comprises hashing header information of the received packet to produce a hash value and indexing a state table using the hash value, wherein the state table includes state information that indicated whether a related packet has already been allowed to reach the CPU.