Serv-Drishti
An interactive visualisation and simulation tool for serverless computing internals. Watch requests traverse your architecture in real-time, analyse cold-start patterns, compare scheduling strategies in Battleground Mode, and export metrics — all in the browser without any backend.
What Serv-Drishti Offers
Interactive Visual Simulation
Watch requests travel through your serverless architecture in real-time with animated flow visualisation and dynamic function pod states.
Comprehensive Analytics
Multiple Chart.js-powered panels for throughput, latency, resource utilisation, cold-start frequency, and detailed metrics tables.
Battleground Mode
Compare routing strategies and placement algorithms side-by-side with synchronised workload injection and head-to-head performance analysis.
Educational Design
Designed for computer science courses, workshops (Distributed Systems, Cloud Computing), and conference demonstrations of serverless concepts.
Highly Configurable
Tune routing strategies, placement algorithms, auto-scaling parameters, concurrency limits, and function warm-up behaviour through a live control panel.
Data Export
Export simulation runs as CSV, JSON, or PNG chart snapshots for inclusion in papers, reports, and slide decks.
Architecture
Core Layer
Chart management, simulation engine, and placement algorithm implementations (round-robin, least-loaded, idle-first, random). All logic runs client-side with no server dependency.
UI Layer
Notification system, live configuration panel, and event handling for user-driven parameter changes during an active simulation run.
Feature Modules
Battleground testing module orchestrates dual-simulation runs with the same seed workload. Export module serialises in-memory simulation state to CSV/JSON/PNG.
Deployment
Static HTML/CSS/JS bundle served via Docker (nginx). Kubernetes-deployable with a single manifest. No external API calls or persistent storage required.
Use Cases
Academic Education
Used in Distributed Systems (COMP90015) and Cloud Computing courses at the University of Melbourne. Gives students hands-on intuition for scheduling and autoscaling trade-offs.
Research Demonstrations
Deployed as a companion demo to the DRe-SCale and Saarthi research papers — shows real-time scheduling decisions under synthetic workload traces at conference presentations.
Enterprise Training
Onboard new team members and demonstrate cold-start impact, concurrency limits, and scale-to-zero behaviour to engineering stakeholders without needing a live cluster.
Technical Implementation
Core Technologies
- Vanilla JavaScript (modular ES6+ architecture, no frameworks)
- Chart.js for performance and analytics visualisations
- Docker (nginx) for containerised deployment
- Kubernetes-deployable with a single manifest file
Simulation Capabilities
- Multiple routing strategies (round-robin, least-loaded, idle-first)
- Auto-scaling simulation with configurable thresholds
- Cold-start modelling with configurable warm-up latency
- Real-time performance metrics and CSV/JSON/PNG export
Publication
Serv-Drishti: An Interactive Serverless Function Request Simulation Engine and Visualiser
Agarwal, S. et al. — IEEE International Telecommunication Networks and Applications Conference
View on IEEE Xplore