US 11,861,352 B2
Smart deployment using graph optimization
Pallavi Baral, Redmond, WA (US); Prateek Punj, Bellevue, WA (US); Yilan Zhang, Redmond, WA (US); Bhuvan Malladihalli Shashidhara, Bellevue, WA (US); Hanyi Xu, Redmond, WA (US); Abhishek Kumar, Kirkland, WA (US); Mayank Meghwanshi, Redmond, WA (US); Sisi Xiong, Redmond, WA (US); Michael Stephenson, Woodinville, WA (US); Avnish Chhabra, Redmond, WA (US); Juan-Arturo Herrera Ortiz, Seattle, WA (US); and Huaming Huang, Redmond, WA (US)
Assigned to Microsoft Technology Licensing, LLC, Redmond, WA (US)
Filed by Microsoft Technology Licensing, LLC, Redmond, WA (US)
Filed on Dec. 29, 2021, as Appl. No. 17/565,191.
Prior Publication US 2023/0205509 A1, Jun. 29, 2023
Int. Cl. G06F 8/65 (2018.01); G06N 20/00 (2019.01)
CPC G06F 8/65 (2013.01) [G06N 20/00 (2019.01)] 20 Claims
OG exemplary drawing
 
1. A system, comprising:
a processor;
a storage having instructions which, when executed by the processor, cause the processor to:
receive a request to deploy an update to target clusters of computers, the target clusters being grouped in multiple levels;
generate multiple layers of graphs including vertices having completed clusters and edges having predicted deployment time values and predicted deployment risk values, the graphs having multiple paths that model deployment plans that indicate orders of the target clusters based on the completed clusters of the vertices in the multiple paths;
finding solutions to shortest path problems presented by the graphs based on the predicted deployment time values and the predicted deployment risk values of the edges and the completed clusters of the vertices in the graphs using at least one optimization algorithm;
outputting deployment plan recommendations corresponding to the solutions;
receiving a selection of a preferred deployment plan from among the deployment plan recommendations; and
deploying the update to the target clusters based on the preferred deployment plan.