Customer complaints in cloud computing can originate from components of the cloud infrastructure platform or from components of third-party software. Further, cloud infrastructure components causing issues can affect customers generally or only customers under certain computing environments or using certain third-party software. Pinpointing the origin of a given customer complaint using targeted testing of components in isolation is computationally infeasible. This disclosure describes techniques that correlate test signals across multiple sources to reliably categorize issues in cloud computing to identify the origin of bad rollouts in a timely and cost-efficient manner. An issue that affects a plurality of workloads can cause test signals generated by the workloads to become correlated. By discovering correlations between signals emitted by distinct workloads, determination can be made of the workloads, customer subsets, computing environments, and third-party software impacted by the issue.

Creative Commons License

Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.