Hyper-scaled control plane managing a multi-tenant key/value datastore.
This is a love-letter to DynamoDB.
Svarm is a large-scale key/value data store. It consists of components that can be used in other projects as well. These include full systems such as the control plane to manage scaling resources in a linear fashion. It also includes sub-components to build other projects out of like the common server code that provides opinionated (Dagger, immutables, Dropwizard) framework extending the great work done by the dropwizard team.
Svarm is for companies that want to rely on their own data centers for a reliable, scalable NoSQL data store. In essence, bringing the cloud back home.
The project is a work-in-progress. For details, see the design.
The control plane itself is designed to be usable for large-scale node management. Details can be found in this doc.
What follows is simply a way to convey what the project does and why you should be interested in it. It is not done. When it is ready for public consumption the work in progress note will be removed.
October 25, 2025. Scottsdale, Arizona: CodeHead Systems is proud to present a new open-source NoSQL key/value datastore that scales linearly, with little to no maintenance. This multi-tenant datastore can quickly consume workloads in the tens of thousands of TPS, support data models with single or multiple partitioning keys, and provides indexing on high-cardinality fields. Companies and research teams can simply provide the hardware, kubernetes instances or docker swarm and start using today either on or off prem.
"Our small lab was able to install svarm within a day and get up and running" says Professor Utonium at PPG Chem LLC. "Real-time data acquisition of our ionic capture array operating at 60Ghz is now possible... resulting in a data transfer rate of well over 25 GB/s. Good thing our fibre optic array with quantum gates is working on our 12 42U racks!"
Download your copy of Apache 2.0 licensed svarm today!
Lol, no, not yet. But if you want to help, please do!
Yes and no. Svarm was intended to provide a generic control plane for any hyper-scaled system. I'm proving it out with a key-value data store. But I do like DynamoDB so why not redo it?
I really like Cassandra. But after years of working with DynamoDB I found it's management and capabilities of working with high-cardinality datasets simpler for the development team. DynamoDB ability to scale also is easier, but of course, AWS hides that complexity with its oncall staff. So building a self-sustaining cluster where the operator only has to add nodes or instances to it seemed like a fun project.
Also, I wanted to build something that could seriously scale. Something that companies can use in their own data center.
The project is designed so many servers can work together to achieve a highly scalable datastore... one that scales linearly. The data nodes themselves can grow and shrink as needed by the collective. 'Swarming' is how I pictur the nodes working together. Svarm is the Swedish word for 'swarm' and was available to register. So... there you go.
I need folks who can write code, documentation, and build websites. This is all open-source and benefits everyone. Come join and be part of the experience.