Skip to main content

Fast Autoscaler

A modular, extensible autoscaling solution for AWS ECS services based on queue metrics.

Overview​

Fast Autoscaler is a production-ready solution that dynamically adjusts ECS task counts based on SQS queue depths and other metrics. Our approach enables precise resource scaling to optimize both performance and cost.

Key Features​

  • Intelligent Queue-based Scaling: Dynamically adjusts ECS service capacity based on SQS queue metrics
  • Priority Scale-up: Override cooldown periods when rapid scaling is needed
  • Configurable Thresholds: Customizable scaling thresholds and cooldown periods
  • Comprehensive Metrics: Support for both visible and in-flight message-based scaling
  • Provider-agnostic Architecture: Extensible design supporting different queue providers (SQS, AMQ, etc.)
  • State Management: S3-based tracking for cooldown periods and scaling history
  • Production-ready Observability: Detailed logging with JSON formatting for CloudWatch integration

Architecture​

The autoscaler follows a modular architecture to support different message queue providers and deployment models:

  • Core Scaling Logic: Centralized scaling decision logic in scaler.py
  • Queue Metrics: Abstracted queue metric collection with provider-specific implementations
  • State Management: S3-based state tracking for scaling events and cooldowns
  • AWS Integration: Lightweight wrapper around AWS services