US 11,989,172 B2
System and method for managing transition between inference models across multiple data processing systems
Ofir Ezrielev, Beer Sheva (IL); Avitan Gefen, Tel Aviv (IL); and Nadav Azaria, Beer Sheva (IL)
Assigned to Dell Products L.P., Round Rock, TX (US)
Filed by Dell Products L.P., Round Rock, TX (US)
Filed on Jul. 12, 2022, as Appl. No. 17/863,027.
Prior Publication US 2024/0020296 A1, Jan. 18, 2024
Int. Cl. G06F 16/23 (2019.01)
CPC G06F 16/2379 (2019.01) 20 Claims
OG exemplary drawing
 
1. A method for executing an inference model across multiple data processing systems that each individually have insufficient computing resources to complete timely execution of the inference model, the method comprising:
obtaining an update for the inference model deployed across the data processing system;
identifying portions of the inference model hosted by the data processing systems and redundancy levels for each of the portions of the inference model;
obtaining an inference model update plan that maintains operation of the inference model during deployment of portions of an updated inference model, the inference model update plan being based on the redundancy levels for each of the portions of the inference model;
obtaining inference model data packages for the data processing systems based on the inference model update plan and the updated inference model;
distributing the inference model data packages to the data processing systems; and
initiating a seamless changeover between execution of the inference model and execution of the updated inference model across the data processing systems using the distributed inference model data packages and a pattern indicated by the inference model update plan.