Processing system and method of detecting congestion in processing system

A method of resolving congestion caused by a specific processing node in a computer system, the computer system comprising:

• a plurality of processing nodes organized in series for sequential processing, each one of the plurality of nodes having a system-unique identifier (SUID),
• a task source for generating a plurality of computer-implemented tasks to be processed by the plurality of nodes, and
• a supervisory computer,
• the method being executable at the supervisory computer,
• the method comprising:
• collecting congestion statistics of the plurality of processing nodes over a pre-determined period of time, the collecting being executed by repeating, over the pre-determined period of time, the steps of:
• receiving, by a given processing node of the plurality of processing nodes, a task packet, the given processing node being a sequential processing node after a previous processing node of the processing pipeline,
• the task packet having:
• (i) a task body indicative of a respective computer-implemented task, and
• (ii) a writable congestion-indicating field;
• processing, by the given processing node, the task body of the task packet;
• determining, by the given processing node, a presence of a respective input queue at the given processing node, the input queue comprising more than one task packet;
• in response to the determining the presence of the respective input queue at the given processing node, updating, by the given processing node, the writable congestion-indicating field with the respective SUID of the given processing node;
• receiving, by a next processing node of the plurality of processing nodes, the task packet, the next processing node being a sequential processing node after the given processing node of the processing pipeline;
• processing, by the next processing node, the task body of the task packet;
• determining, by the next processing node, a presence of a respective input queue at the next processing node;
• and
• in response to the determining the presence of the respective input queue at the next processing node, updating, by the next processing node, the writable congestion-indicating field with the respective SUID of the next processing node;
• reading, by the supervisory computer, the writable congestion-indicating field of the task packet having exited the processing pipeline, thereby identifying a last-written SUID having been written into the writable congestion-indicating field, the last-written SUID being indicative of a most-downstream processing node of the plurality of processing nodes causing congestion in the processing pipeline;
• and
• recording the last-written SUID to the congestion statistics;
• identifying, by the supervisory computer, a most common last-written SUID recorded to the congestion statistics;
• identifying the specific node associated with the most common last-written SUID;
• and
• causing at least one parameter associated with the specific node to be changed.

A computer system comprising:

• a task source for generating a stream of computer-implemented tasks to be processed by the computer system;
• and
• a plurality of processing nodes arranged in a sequence, the plurality of processing nodes forming a processing pipeline for processing the stream of computer-implemented tasks, each one of the plurality of processing nodes having a respective system-unique identifier (SUID), each given processing node of the plurality of processing nodes being configured to perform:
• receiving a task packet from a sequentially previous processing node of the processing pipeline, the task packet having:
• (i) a task body indicative of a respective computer-implemented task, and
• (ii) a writable congestion-indicating field;
• processing the task body of the task packet;
• determining a presence of a respective input queue at the given processing node, the input queue comprising more than one task packet;
• and
• in response to the determining the presence of the respective input queue at the given processing node, updating the writable congestion-indicating field with the respective SUID of the given processing node.

A method of detecting congestion in a computer system, the computer system having:

• (i) a plurality of processing nodes arranged in a sequence and forming a processing pipeline for processing a stream of computer-implemented tasks generated by a task source,
• (ii) each one of the plurality of processing nodes having a respective system-unique identifier (SUID),
• the method being executable at the computer system, the method comprising:
• collecting congestion statistics of the plurality of processing nodes over a pre-determined period of time, the collecting being executed by repeating, over the pre-determined period of time, the steps of:
• receiving, by a given processing node of the plurality of processing nodes, a task packet, the given processing node being a sequential processing node after a previous processing node of the processing pipeline,
• the task packet having:
• (iii) a task body indicative of a respective computer-implemented task, and
• (iv) a writable congestion-indicating field;
• processing, by the given processing node, the task body of the task packet;
• determining, by the given processing node, a presence of a respective input queue at the given processing node, the input queue comprising more than one task packet;
• in response to the determining the presence of the respective input queue at the given processing node, updating, by the given processing node, the writable congestion-indicating field with the respective SUID of the given processing node;
• receiving, by a next processing node of the plurality of processing nodes, the task packet, the next processing node being a sequential processing node after the given processing node of the processing pipeline;
• processing, by the next processing node, the task body of the task packet;
• determining, by the next processing node, a presence of a respective input queue at the next processing node;
• and
• in response to the determining the presence of the respective input queue at the next processing node, updating, by the next processing node, the writable congestion-indicating field with the respective SUID of the next processing node;
• reading, by the computer system, the writable congestion-indicating field of the task packet having exited the processing pipeline, thereby identifying a last-written SUID having been written into the writable congestion-indicating field, the last-written SUID being indicative of a most-downstream processing node of the plurality of processing nodes causing congestion in the processing pipeline;
• and
• recording the last-written SUID to the congestion statistics;
• identifying, by the computer system, a most common last-written SUID recorded to the congestion statistics;
• identifying the specific node associated with the most common last-written SUID;
• and
• causing at least one parameter associated with the specific node to be changed.