Skip to main content

Architecture

Oodle is a fast, fully managed serverless metrics observability platform that makes it radically simple, and cost-effective to monitor all your infra/apps. The core principle of Oodle's architecture is the separation of storage and compute. This allows for independent scaling of both components, resulting in high performance at a significantly lower cost.

Oodle Architecture Diagram

Compute Layer

Traditional systems over-provision compute resources to handle peak loads, often leading to unnecessary costs. While some systems use auto-scaling to manage this, auto-scaling isn't fast enough for real-time observability needs. The compute layer of Oodle's architecture leverages serverless functions, specifically AWS Lambda, to provide on-demand and instantly scalable compute resources. This approach enables the parallelization of queries across serverless instances, significantly enhancing performance. By utilizing serverless technology, Oodle reduces costs as customers only pay for actual usage, making it a highly efficient and cost-effective solution for metrics observability.

Storage Layer

Traditional metrics observability systems use SSDs with 2 or 3-way replication, which is both costly and redundant. Oodle's storage layer utilizes object storage, specifically Amazon S3, instead of traditional SSDs. This approach is significantly cheaper, offering a 10x cost reduction compared to SSD-based storage. It allows for long-term retention of high-fidelity data without the need for downsampling. The system employs custom formatting to store indexes alongside data on S3, optimizing retrieval efficiency. Furthermore, S3 provides 99.999999999% durability, effectively protecting against data corruption and loss. This robust storage solution forms a crucial part of Oodle's scalable and cost-effective architecture.

Networking and Availability

Oodle's architecture is designed with Availability Zone (AZ) awareness to minimize egress costs. This approach ensures that data scraping and ingestion processes occur within the same availability zone, optimizing data transfer efficiency. Additionally, the system is engineered for resilience, capable of maintaining operations even in the event of availability zone failures. This robust design enhances both cost-effectiveness and reliability of the platform.

Integration and Compatibility

Oodle's architecture is designed with integration and compatibility in mind. The platform supports open-source protocols such as OpenTelemetry and Prometheus Remote Write, ensuring seamless integration with existing monitoring setups. Additionally, Oodle is compatible with popular visualization tools like Grafana, allowing users to leverage familiar interfaces for data analysis and dashboard creation. This comprehensive approach to integration and compatibility enhances the platform's versatility and user-friendliness.

This architecture enables Oodle's platform to handle more than 1 billion time series per hour while maintaining cost-efficiency and high performance. The system is designed to address common challenges in metrics observability, such as scaling issues, high costs, and performance bottlenecks.

Further Reading

For a more in-depth exploration of Oodle's architecture you can read our detailed blog post: Building a High-Performance, Low-Cost Metrics Observability System.