Mingming Liu
Graph Neural Networks for Micro Service Based Cloud Application
Recently, there has been a significant shift towards the adoption of microservice architectures for cloud-based applications. The key idea of these architectures is to allow a group of small, independent, and scalable functional units working together to achieve common goals through efficient communication protocols and mechanisms facilitated in cloud networks. As applications increasingly adopt the microservice architectures, it becomes practically important to have intelligent resource provisioning algorithms in place. The primary objective of these intelligent algorithms is to dynamically allocate cloud resources such as CPU, memory to different microservices as their demands change over time, which can help improve efficiency and reduce operational costs.
The design of an intelligent resource provisioning algorithm often considers various factors including the workload pattern of each microservice, the available resources in the cloud environment, and the service level agreements (SLAs) that the applications must meet. Machine learning algorithms have become increasingly popular in recent years for designing these algorithms for microservice-based applications. By analyzing historical data available on cloud-based monitoring platforms, these algorithms are capable of forecasting future resource requirements and performance of applications to the next level
In this talk, I will present some recent advances in using machine learning techniques for the design of these algorithms, starting from basic rule-based methods to recent graph-based learning methods.
The design of an intelligent resource provisioning algorithm often considers various factors including the workload pattern of each microservice, the available resources in the cloud environment, and the service level agreements (SLAs) that the applications must meet. Machine learning algorithms have become increasingly popular in recent years for designing these algorithms for microservice-based applications. By analyzing historical data available on cloud-based monitoring platforms, these algorithms are capable of forecasting future resource requirements and performance of applications to the next level
In this talk, I will present some recent advances in using machine learning techniques for the design of these algorithms, starting from basic rule-based methods to recent graph-based learning methods.
back to overview
Watch Recording
