An embedded database for serverless and edge runtimes, storing data as Parquet on S3 with stateless compute.
Tonbo is an embedded database designed specifically for serverless and edge computing environments. It bridges the gap between stateless compute and persistent data by storing data as Parquet files on object storage like S3 and coordinating through a manifest, eliminating the need to manage a database server.
Developers building serverless or edge applications that need a durable state layer without running a database, such as those using Cloudflare Workers, Deno, or AWS Lambda. It also targets developers who need to embed a lightweight MVCC + Parquet storage engine inside their own data infrastructure.
Developers choose Tonbo because it provides a fully async, serverless-native database with no server to manage, using open Parquet files on S3 for vendor-agnostic data storage. Its unique selling point is enabling stateless compute with MVCC transactions and snapshot isolation directly on object storage.
Tonbo is an embedded database for serverless and edge runtimes.
Open-Awesome is built by the community, for the community. Submit a project, suggest an awesome list, or help improve the catalog on GitHub.
Fully async engine built from the ground up for serverless and edge environments, as highlighted in the key features, ensuring optimal performance in stateless compute contexts.
Eliminates database servers by storing data as Parquet on S3 and coordinating via a manifest, reducing operational overhead and aligning with serverless philosophy per the README.
Uses standard Parquet files readable by any tool like DataFusion or Spark, preventing vendor lock-in and enabling easy data portability, as emphasized in the philosophy section.
Provides snapshot isolation and consistent point-in-time reads through MVCC, enabling reliable transactions in distributed, stateless environments without a central coordinator.
Project is in alpha with APIs subject to change, as noted in the status section, making it risky for production workloads and requiring frequent updates.
Lacks filter pushdown and complex query support like joins, with these features marked as in-progress, restricting use to simpler scan-and-filter operations.
Relies on object storage like S3 for all data operations, which can introduce higher latency and costs compared to in-memory or local-storage databases, affecting real-time performance.
Primarily supports Rust with Python and JavaScript bindings still in progress, limiting adoption for teams using other languages and increasing the learning curve.