Method and system for information exchange utilizing an asynchronous persistent store protocol

A method for information exchange through a workload manager managing units of data in need of processing, the workload manager in communication with a client and at least one compute node in a distributed computing environment, the method comprising:

• receiving by the workload manager a workload unit of data in need of processing from the client, wherein the client expects a result from the at least one compute node based on and in response to the workload unit;
• initiating by the workload manager a persistent storage of the workload unit received from the client;
• without waiting for the initiated storage of the workload unit received from the client to complete, sending by the workload manager the workload unit received from the client to the at least one compute node;
• receiving the result from the at least one compute node;
• aborting by the workload manager the initiated storage of the workload unit received from the client if, after the receiving the result from the at least one compute node, the initiated storage of the workload unit received from the client is not complete;
• initiating by the workload manager a persistent storage of the result received from the at least one compute node;
• without waiting for the initiated storage of the result received from the at least one compute node to complete, sending by the workload manager the result received from the at least one compute node to the client;
• receiving by the workload manager an acknowledgement from the client of receipt of the result;
• and
• aborting by the workload manager the initiated storage of the result received from the at least one compute node if, after the receiving the acknowledgement from the client of the result sent to the client, the initiated storage of the result received from the at least one compute node is not complete,
• upon experiencing a failure by the workload manager after the result has been stored and before the result has been received by the client, determining that the workload manager has recovered from the failure to retrieve the stored result and continue the information exchange in a manner transparent to both the client and the at least one computer node;
• and
• upon experiencing a failure by the client after the client has received the result and before the client has sent an acknowledgement, determining that the client has recovered from the failure and sending the result, by the workload manager, to the client, wherein a new compute node may be selected to receive the workload unit if the workload manager or the client experience a failure,
• wherein the manager determines the at least one compute node upon receiving the workload unit from the client, wherein the at least one compute node is not pre-determined by the client.

A system for information exchange across a computing environment between a source and a target, the system comprising:

• a persistent store device;
• and
• a manager of data in need of processing, the manager in communication with the persistent store device, the source, and the target, wherein the manager is configured to:
• receive a data input from the source, wherein the source expects a data output from the target based on and in response to the data input,
• initiate a storage in the persistent store device of the data input received from the source,
• send the data input received from the source to the target without waiting for the initiated storage of the data input received from the source to complete,
• receive the data output from the target,
• abort the initiated storage of the data input received from the source if, after the data output is received from the target, the initiated storage of the data input received from the source is not complete, and
• initiate a storage in the persistent store device of the data output received from the target,
• send the data output received from the target to the source without waiting for the initiated storage of the data output received from the target to complete,
• receive an acknowledgement from the source of receipt of the data output, and
• abort the initiated storage of the data output received from the target if, after receiving the acknowledgement from the source of receipt of the data output, the initiated storage of the data output received from the target is not complete,
• upon the source experiencing a failure by the source after the source has received the data output and before the source has sent an acknowledgment, determining that the source has recovered from the failure and sending the output by the manager to the source, wherein a new target may be selected to receive the data input if the manager or the source experience a failure;
• and
• upon experiencing a failure by the manager after the data output has been stored and before the data output has been received by the source, determining that the manager has recovered from the failure to retrieve the stored data output and continue the information exchange in a manner transparent to both the source and the target,
• wherein the manager determines targets upon receiving the data input from the source, wherein targets are not pre-determined by the source.

A system for information exchange between a client and at least one compute node in a distributed computing environment, the system comprising:

• a persistent store device;
• and
• a workload manager managing units of data in need of processing, the workload manager in communication with the persistent store device, the client, and the at least one compute node, wherein the workload manager is configured to:
• receive a workload unit of data in need of processing from the client, wherein the client expects a result from the at least one compute node based on and in response to the workload unit,
• initiate a storage in the persistent store device of the workload unit received from the client,
• send the workload unit received from the client to the at least one compute node without waiting for the initiated storage of the workload unit received from the client to complete,
• receive the result from the at least one compute node, and
• abort the initiated storage of the workload unit received from the client if, after the result is received from the at least one compute node, the initiated storage of the workload unit received from the client is not complete, and
• initiate a storage in the persistent store device of the result received from the at least one compute node,
• send the result received from the at least one compute node to the client without waiting for the initiated storage of the result received from the at least one compute node to complete,
• receive an acknowledgement from the client of receipt of the result, and
• abort the initiated storage of the result received from the at least one compute node if, after receiving the acknowledgement from the client of receipt of the result, the initiated storage of the result received from the at least one compute node is not complete,
• upon the client experiencing a failure after the client has received the result and before the client has sent an acknowledgment determining that the client has recovered from the failure and sending the result, by the workload manager, to the client, wherein a new compute node may be selected to receive the workload unit if the workload manager or the client experience a failure;
• and
• upon experiencing a failure by the workload manager after the result has been stored and before the result has been received by the client, determining that the workload manager has recovered from the failure to retrieve the stored result and continue the information exchange in a manner transparent to both the client and the at least one computer node,
• wherein the manager determines the at least one compute node upon receiving the workload unit from the client, wherein the at least one compute node is not pre-determined by the client.