Scalable resource management in distributed environment

Abstract

A method may include receiving a packet in a network device, selecting one of a group of ingress buffers, where each ingress buffer is associated with a different one of a group of processors, distributing the packet to the selected ingress buffer; and scheduling the packet, based on a congestion state of a queue in an egress buffer associated with the packet, to be processed by the processor associated with the selected ingress buffer to provide a network service

1 - 20 . (canceled) 
     
     
         21 . A method comprising:
 receiving a packet in a network device;   selecting one of a plurality of ingress buffers;   distributing the packet to the selected ingress buffer;   receiving, from an egress buffer associated with the packet, a congestion state of a queue in the egress buffer;   determining whether to schedule the packet to be processed by a processor associated with the selected ingress buffer based on whether the received congestion state indicates that the queue in the egress buffer is congested and the congestion state was received not more than a threshold amount of time earlier; and   processing, by the processor associated with the selected ingress buffer, the packet when the packet is scheduled to be processed by the processor.   
     
     
         22 . The method of  claim 21 , where determining whether to schedule the packet to be processed further includes not permitting the processing of the packet when the received congestion state indicates that the queue in the egress buffer is congested and the congestion state was received not more than a threshold amount of time earlier. 
     
     
         23 . The method of  claim 21 , where determining whether to schedule the packet further includes:
 determining to schedule the packet to be processed when the congestion state was received more than a threshold time earlier and the received congestion state indicates that the queue in the egress buffer is congested.   
     
     
         24 . The method of  claim 21 , where selecting the one of the plurality of ingress buffers includes:
 classifying the packet based on one or more of a source address, a destination address, a source port, a destination port, a protocol, or an application associated with the packet; and   selecting the one of the plurality of ingress buffers based on the classifying.   
     
     
         25 . The method of  claim 24 , where classifying the packet includes determining whether the packet belongs to a class, a flow, or a session; and where distributing the packet includes distributing the packet based on the determination of whether the packet belongs to the class, the flow, or the session. 
     
     
         26 . The method of  claim 21 , further comprising:
 determining a length of the queue in the egress buffer;   indicating the congestion state as congested when the length is greater than a first threshold; and   indicating the congestion state as not congested when the length is less than a second threshold.   
     
     
         27 . The method of  claim 26 , further comprising receiving the congestion state of the queue in the egress buffer when the state changes or periodically when the state does not change. 
     
     
         28 . A network device comprising:
 one or more processors to:
 receive a packet; 
 select one of a plurality of ingress buffers; 
 distribute the packet to the selected ingress buffer; 
 receive, from an egress buffer associated with the packet, a congestion state of a queue in the egress buffer; 
 determine whether to schedule the packet to be processed by a processor associated with the selected ingress buffer based on whether the received congestion state indicates that the queue in the egress buffer is congested and the congestion state was received not more than a threshold amount of time earlier; and 
 process the packet when the packet is scheduled to be processed. 
   
     
     
         29 . The network device of  claim 28 , where, when determining whether to schedule the packet to be processed, the one or more processors are further to not schedule the packet to be processed when the received congestion state indicates that the queue in the egress buffer is congested and the congestion state was received not more than a threshold amount of time earlier. 
     
     
         30 . The network device of  claim 28 , where, when determining whether to schedule the packet, the one or more processors are further to:
 determine to schedule the packet to be processed when the congestion state was received more than a threshold time earlier and the received congestion state indicates that the queue in the egress buffer is congested.   
     
     
         31 . The network device of  claim 28 , where, when selecting the one of the plurality of ingress buffers, the one or more processors are further to:
 classify the packet based on one or more of a source address, a destination address, a source port, a destination port, a protocol, or an application associated with the packet; and   select the one of the plurality of ingress buffers based on the classifying.   
     
     
         32 . The network device of  claim 31 , where, when classifying the packet, the one or more processors are further to determine whether the packet belongs to a class, a flow, or a session; and
 where, when distributing the packet, the one or more processors are further to distribute the packet based on the determination whether the packet belongs to the class, the flow, or the session.   
     
     
         33 . The network device of  claim 28 , where the one or more processors are further to:
 determine a length of the queue in the egress buffer;   indicate the congestion state as congested when the length is greater than a first threshold; and   indicate the congestion state as not congested when the length is less than a second threshold.   
     
     
         34 . The network device of  claim 33 , where the one or more processors are further to receive the congestion state of the queue in the egress buffer when the state changes or periodically when the state does not change. 
     
     
         35 . The network device of  claim 33 , where, when determining whether to schedule the packet to be processed, the one or more processors are further to:
 hold the packet in the selected one of the plurality of ingress buffers without scheduling the packet for processing when the congestion state was received not more than the threshold time earlier and the received congestion state indicates that the queue in the egress buffer is congested.   
     
     
         36 . The network device of  claim 28 , where the one or more processors are further to:
 output the processed packet to a destination of the processed packet, where the destination is another network device.   
     
     
         37 . A network device comprising:
 a receiver to receive a packet;   a plurality of memories, each of the plurality of memories storing a different ingress buffer and at least one of the plurality of memories storing an egress buffer, associated with a plurality of the different ingress buffers, the egress buffer including a queue associated with the packet; and   a plurality of processors, each processor associated with a different ingress buffer,   a first processor, of the plurality of processors, to:
 select one of the plurality of different ingress buffers, and 
 distribute the packet to the selected ingress buffer, 
   a processor associated with the selected ingress buffer to:
 receive, from an egress buffer associated with the packet, a congestion state of a queue in the egress buffer; 
 determine whether to schedule the packet to be processed based on whether the received congestion state indicates that the queue in the egress buffer is congested and the congestion state was received not more than a threshold amount of time earlier; and 
 process the packet when the packet is scheduled to be processed. 
   
     
     
         38 . The network device of  claim 37 , where, when determining whether to schedule the packet to be processed, the processor associated with the selected ingress buffer is further to not schedule the packet to be processed when the received congestion state indicates that the queue in the egress buffer is congested and the congestion state was received not more than a threshold amount of time earlier. 
     
     
         39 . The network device of  claim 37 , where, when determining whether to schedule the packet to be processed, the processor associated with the selected ingress buffer is further to:
 determine to schedule the packet to be processed when the congestion state was received more than a threshold time earlier and the received congestion state indicates that the queue in the egress buffer is congested.   
     
     
         40 . The network device of  claim 37 , where, when selecting the one of the plurality of ingress buffers, the first processor is further to:
 classify the packet based on one or more of a source address, a destination address, a source port, a destination port, a protocol, or an application associated with the packet; and   select the one of the plurality of ingress buffers based on the classifying.