Serv-Drishti: An Interactive Serverless Function Request Simulation Engine and Visualiser

TL;DR

Serv-Drishti addresses the opacity of serverless platforms by offering an interactive, open-source end-to-end simulation and visualisation tool for Function-as-a-Service workflows. Built as a discrete-event, request-driven model, it simulates the journey from API Gateway through a Request Dispatcher to Compute Nodes hosting Function Instances, supporting multiple routing and placement strategies, dynamic scaling, and a robust failure module. Key contributions include the visualisation-centric architecture, a battleground feature for direct strategy comparisons, extensive data export capabilities, and extensibility through modular interfaces for custom logic. The platform enables educators and researchers to analyze latency, throughput, resource utilisation, and costs in a risk-free environment, advancing understanding and design decisions for serverless architectures; future work will add predictive scaling, network-latency visualization, richer failure modes, and DAG-based workflows, with the project available under the Apache 2.0 license on GitHub.

Abstract

The rapid adoption of serverless computing necessitates a deeper understanding of its underlying operational mechanics, particularly concerning request routing, cold starts, function scaling, and resource management. This paper presents Serv-Drishti, an interactive, open-source simulation tool designed to demystify these complex behaviours. Serv-Drishti simulates and visualises the journey of a request through a representative serverless platform, from the API Gateway and intelligent Request Dispatcher to dynamic Function Instances on resource-constrained Compute Nodes. Unlike simple simulators, Serv-Drishti provides a robust framework for comparative analysis. It features configurable platform parameters, multiple request routing and function placement strategies, and a comprehensive failure simulation module. This allows users to not only observe but also rigorously analyse system responses under various loads and fault conditions. The tool generates real-time performance graphs and provides detailed data exports, establishing it as a valuable resource for research, education, and the design analysis of serverless architectures.

Figures (9)

• Figure 1: Request Flow across Serv-Drishti Components
• Figure 2: Different Placement Algorithm Performance Comparison
• Figure 3: Function Scaling Over Simulation
• Figure 4: Impact of Request Queue and Cold Start on Function Request Performance
• Figure 5: A Failure Simulation through Fail Node