System and method for supporting efficient load-balancing in a high performance computing (hpc) environment

A method for supporting efficient load balancing among a plurality of switches and a plurality of end nodes arranged in a tree topology in a network environment, the method comprising:

• sorting the plurality of end nodes, the plurality of end nodes being on one more of the plurality of switches, wherein the plurality of end nodes are sorted in a decreasing order of a receive weight;
• and
• routing, in the decreasing order of receive weights, the plurality of end nodes, wherein the routing comprises selecting at least one down-going port and at least one up-going port;
• wherein selecting the at least one down-going port comprises
• comparing a plurality of down-going ports and selecting the down-going port having a least accumulated downward weight, and
• in response to two or more down-going ports having the least accumulated downward weight and the least accumulated upward weight, comparing the two or more down-going ports having the least accumulated downward weight and the least accumulated upward weight, and selecting from the two or more down-going ports having the least accumulated downward weight and the least accumulated upward weight the down-going port having a smallest global unique identifier.

A system for supporting efficient load balancing among a plurality of switches and a plurality of end nodes arranged in a tree topology in a network environment, the system comprising:

• one or more microprocessors;
• and
• a processor, running on the one or more microprocessors, wherein the processor operates to perform steps comprising:
• sorting the plurality of end nodes, the plurality of end nodes being on one more of the plurality of switches, wherein the plurality of end nodes are sorted in a decreasing order of a receive weight;
• and
• routing, in the decreasing order of receive weights, the plurality of end nodes, wherein the routing comprises selecting at least one down-going port and at least one up-going port;
• wherein selecting the at least one down-going port comprises
• comparing a plurality of down-going ports and selecting the down-going port having a least accumulated downward weight, and
• in response to two or more down-going ports having the least accumulated downward weight;
• comparing the two or more down-going ports having the least accumulated downward weight and selecting from the two or more down-going ports having the least accumulated downward weight the down-going port having a least accumulated upward weight.

A system for supporting efficient load balancing among a plurality of switches and a plurality of end nodes arranged in a tree topology in a network environment, the method comprising:

• one or more microprocessors;
• and
• a processor, running on the one or more microprocessors, wherein the processor operates to perform steps comprising:
• sorting the plurality of end nodes, the plurality of end nodes being on one more of the plurality of switches, wherein the plurality of end nodes are sorted in a decreasing order of a receive weight;
• and
• routing, in the decreasing order of receive weights, the plurality of end nodes, wherein the routing comprises selecting at least one down-going port and at least one up-going port;
• wherein selecting the at least one down-going port comprises
• comparing a plurality of down-going ports and selecting the down-going port having a least accumulated downward weight, and
• in response to two or more down-going ports having the least accumulated downward weight and the least accumulated upward weight, comparing the two or more down-going ports having the least accumulated downward weight and the least accumulated upward weight, and selecting from the two or more down-going ports having the least accumulated downward weight and the least accumulated upward weight the down-going port having a smallest global unique identifier.

A non-transitory machine readable storage medium having instructions stored thereon for supporting efficient load balancing among a plurality of switches and a plurality of end nodes arranged in a tree topology in a network environment that when executed cause a system to perform steps comprising:

• sorting the plurality of end nodes, the plurality of end nodes being on one more of the plurality of switches, wherein the plurality of end nodes are sorted in a decreasing order of a receive weight;
• and
• routing, in the decreasing order of receive weights, the plurality of end nodes, wherein the routing comprises selecting at least one down-going port and at least one up-going port;
• wherein selecting the at least one down-going port comprises
• comparing a plurality of down-going ports and selecting the down-going port having a least accumulated downward weight, and
• in response to two or more down-going ports having the least accumulated downward weight;
• comparing the two or more down-going ports having the least accumulated downward weight and selecting from the two or more down-going ports having the least accumulated downward weight the down-going port having a least accumulated upward weight.

A non-transitory machine readable storage medium having instructions stored thereon for supporting efficient load balancing among a plurality of switches and a plurality of end nodes arranged in a tree topology in a network environment that when executed cause a system to perform steps comprising:

• sorting the plurality of end nodes, the plurality of end nodes being on one more of the plurality of switches, wherein the plurality of end nodes are sorted in a decreasing order of a receive weight;
• and
• routing, in the decreasing order of receive weights, the plurality of end nodes, wherein the routing comprises selecting at least one down-going port and at least one up-going port;
• wherein selecting the at least one down-going port comprises
• comparing a plurality of down-going ports and selecting the down-going port having a least accumulated downward weight, and
• in response to two or more down-going ports having the least accumulated downward weight and the least accumulated upward weight, comparing the two or more down-going ports having the least accumulated downward weight and the least accumulated upward weight, and selecting from the two or more down-going ports having the least accumulated downward weight and the least accumulated upward weight the down-going port having a smallest global unique identifier.