Resource Provider Determination in Cloud Environments

Hybrid cloud infrastructure is composed of two or more clouds that inter-operate or federate through technology. The resources required for computing including storage, network, memory etc. can be requested and leased out in a hybrid, and or multi-cloud environment. These cloud resources are required only to satisfy particular large workload needs and can be released afterwards. There is a need to dynamically identify the required resources shortage and generate appropriate lease requests to provision identified resources from another cloud for the duration of the task in need. This can be very helpful in model training use cases. We address this challenge by proposing a system which can execute resource generation requests from appropriate providers in a hybrid cloud environment, while adhering to the constraints such as security policies. In our method, the resource borrowing or lease requests are submitted to another cloud, and a matching is generated while honoring the security policies of both sides. We approach this problem using natural language processing technologies and reinforcement learning. In this paper, we talk about the existing work in this area, the problem which we intend to solve, and the related future work in this direction.