Amazon Technologies, Inc.

Siyu Wang

Letian Feng

Malcolm Featonby

Shaokang Ni

Zachary Daniel Casper

Nathaniel Baird Jones

Leslie Johann Lamprecht

Qijia Chen

Chia-Yu Kao

Systems and methods for implementing a right-sizing service that recommends and automatically implements resource sizing recommendations for compute clusters executing applications is disclosed. Performance metrics for compute nodes executing the application may be gathered and normalized, with anomalous nodes excluded. Then, respective time series data for the compute nodes may be temporally aligned to identify phases of execution of the application and resource sizes may be estimated for the various phases. A resource sizing recommendation may then be made based in part on this estimated resource sizes. The sizing recommendation may further integrate with other sizing recommendations and be used to manually or automatically size a compute cluster or selected among a number of pre-sized compute clusters, in various embodiments.

Resource right-sizing for compute clusters

Systems and methods for implementing a right-sizing service that recommends and automatically implements sizing recommendations for nodes and resources of compute clusters executing applications is disclosed. Performance metrics for compute nodes executing the application may be gathered and normalized with anomalous nodes excluded. Then, respective time series data for the compute nodes may be temporally aligned to identify phases of execution of the application and resource sizes may be estimated for the various phases. A sizing recommendation may then be made based in part on this estimated resource sizes. The sizing recommendation may further integrate with other sizing recommendations and be used to manually or automatically size a compute cluster or selected among a number of pre-sized compute clusters, in various embodiments.

Cluster right-sizing for cloud-based applications

Techniques for a cloud-based workload optimization service to identify customer workloads that are optimized to run on burstable instance types. The techniques include identifying workloads that are successfully running on burstable instance types, and using historical-utilization data for those workloads to train classification models. The optimization service can extract feature data from the historical-utilization data, where the feature data represents utilization characteristics that are indicative of burstable workloads. The feature data is then used to train classification models to receive utilization data for candidate workloads, and determine whether the candidate workloads would be optimized for burstable instance types. The optimization service can then migrate suitable workloads to burstable instance types, and/or provide users with recommendations that their workloads are optimized or suitable for burstable instance types.

Classifying workloads for burstable compute platforms

Growing community of inventors

Seattle, WA, United States of America

Siyu Wang