Evaluation of Integrated Frameworks for Optimizing QoS in Serverless Computing

Serverless computing is an emerging cloud deployment model where developers can concentrate on developing application logic without worrying about the underlying architecture. It is similar to the platform as a service (PaaS) but at the functional level. Applications are usually deployed in the form of a set of functions independently and each function may be executed at separate servers thus also named as function as a service (FaaS). Serverless at the edge can handle thousands of concurrent functions invocations to process various kinds of events generated from resources like database, system logs, and other storage units, etc. A number of serverless frameworks like Openfaas, Openwhisk, Microsoft Azure, Amazon AWS allow dynamic scaling to handle the parallel request of stateless functions from the client-side. A separate container manager may be provisioned to handle distributed load for data processing. In this paper, we have evaluated the performance of serverless frameworks for parallel loads in terms of response time and throughput. In this paper, we have shown that the serverless framework is suitable for handling dynamic applications that can be executed on a number of stateless functions. An extensive comparison of the performance of serverless frameworks in handling concurrent invocations in terms of response time and throughput is also presented. It has been observed that Openwhisk is found to be the better serverless framework in terms of elasticity and scalability.