Cloud computing has overgrown, leading to the transformation of workloads from on-premises server rooms into public cloud environments. This change in the workflow has benefited developers by shifting the responsibility of server and infrastructure management to a cloud service provider. Along with these services, a new emerging paradigm for cloud computing is introduced called serverless computing, where the user does not need to manage any server infrastructure. FaaS is a serverless computing model, that allows developers to deploy individual functions to the cloud, where cloud providers take care of resource management tasks such as resource provisioning, deployment, and auto-scaling. This paper focuses on choosing the most favorable resource size for the AWS lambda function for optimal cost by using a reference table. It allows developers to take a reference while choosing the memory size for the lambda function. The output was evaluated on 10 different serverless functions on AWS. Initially, these functions were set up using the default memory size, and during the evaluation, the referred memory size was chosen. Based on the evaluation of test functions, it selects the optimal memory size for 80.0% of the serverless functions, which results in an average speedup of 142.26% while also decreasing average costs by 10.57%.

cloud computing

function as a service

lambda

lambda function optimization

dynamic profiling

event-driven computing

lambda cost-performance tradeoff