Serverless Function Monitor

Part of the worlds-biggest-software-project initiative.

An AI-native, OpenTelemetry-based observability platform for serverless functions that makes cold start tracking, cost-per-invocation attribution, and performance analytics affordable and vendor-neutral.

Serverless Function Monitor is an open-source monitoring system for AWS Lambda, Google Cloud Functions, and Azure Functions. It is built for cloud engineers, backend developers, FinOps practitioners, and SREs who need deep visibility into invocation performance and cost without paying enterprise observability prices.

Why Serverless Function Monitor?

Incumbent platforms are expensive. Datadog Serverless and Lumigo typically cost $500–$1,200/mo, with pricing models that scale unpredictably as function volume grows.
Native tooling is thin. AWS CloudWatch is zero-setup but offers limited visualisation and expensive log storage, requiring significant engineering to build useful dashboards.
Cold start costs jumped sharply. AWS's August 2025 INIT phase billing change increased cold start costs by up to 22x, making cost visibility urgent for Lambda users.
Multi-cloud serverless monitoring is unsolved. Most current tools focus on AWS Lambda; unified Lambda + Cloud Run + Azure Functions visibility is a documented gap.
The market has consolidated into full-stack platforms. Pure-play serverless specialists (Epsagon, Thundra) have been acquired or absorbed, leaving room for an open-source, OTel-native alternative.

Key Features

Invocation & Performance Monitoring

Function invocation tracking and metadata
Execution time and duration metrics
Cold start detection and tracking
Memory usage monitoring
Invocation count and frequency
Duration and latency percentiles

Errors, Logs & Alerting

Error and exception logging
Log aggregation and search
Basic alerting rules
Automatic error categorization (v1.1)
Timeout detection and analysis (v1.1)

Cost & Optimisation

Cost per invocation calculation (v1.1)
Cost trending and forecasting (v1.1)
Automatic cost optimization recommendations (backlog)
Reserved capacity recommendation (backlog)

Tracing & Architecture

Distributed tracing across functions (v1.1)
Function dependency mapping (v1.1)
Concurrent execution tracking (v1.1)
Serverless architecture visualization (backlog)
Function version comparison (backlog)

Multi-Cloud (Backlog)

Multi-vendor monitoring across AWS, GCP, and Azure
Automatic performance regression detection
Integration with incident management

AI-Native Advantage

AI is used to recommend optimal memory configurations by analysing cost-vs-duration curves across invocation history, replacing manual Lambda Power Tuning runs. It detects cold start regressions after deployments and correlates them to dependency size changes, runtime updates, or initialisation code patterns. Cost attribution reports map invocation costs to business workflows without manual tagging, and natural-language incident investigation answers questions like "why did my order-processor function timeout at 2pm on Tuesday?" with a correlated trace, log, and metric timeline.

Tech Stack & Deployment

The project targets OpenTelemetry-native instrumentation: OTel Lambda layers and the OTel Lambda extension for telemetry export, OTLP as the wire protocol, and the AWS Lambda Extensions API for low-overhead agent attachment. CloudWatch Embedded Metric Format (EMF) is used for low-cost custom metrics. The system is designed for self-hosted and cloud deployment modes, aligning with the FinOps Framework for cost attribution and showback/chargeback.

Market Context

The serverless monitoring market sits within the broader cloud monitoring space, estimated at USD 4.2 billion in 2025, with AWS Lambda alone running hundreds of billions of invocations per month. Incumbents typically charge $500–$1,200/mo (Datadog, Lumigo), while Atatus starts at $49/mo and CloudWatch is consumption-based. Primary buyers are cloud engineers managing Lambda costs, backend developers debugging cold start latency, FinOps practitioners doing function-level cost attribution, and SREs managing SLOs for event-driven architectures. Teams report 40% MTTR reduction and 30% lower operational costs with comprehensive serverless monitoring.

Project Status

This project is in the research and specification phase.
Contributions, feedback, and domain expertise are welcome.

Contributing

We welcome contributions from developers, domain experts, and potential users. See CONTRIBUTING.md for guidelines.

Important: All contributions must be your own original work or clearly attributed open-source material with a compatible licence. Copyright infringement and licence violations will not be tolerated and will result in immediate removal of the offending contribution. If you are unsure whether a piece of code, text, or other material is safe to contribute, open an issue and ask before submitting.

Licence

Licence to be determined. See discussion for context.