Deployment models, what to expect and timelines
Here's an update on the development of The Things Network Stack V3, deployment models, V3 API and timelines.
First, I would like to thank you for your patience. While we serve and support your business with V2, we are aware of its shortcomings. We're happy to get closer and closer to migrating you to V3.
Second, V3 is going to be fantastic. It's been a big investment. We've been working on it for 14 months with 6 engineers. We started writing from scratch, taking all lessons learned into account from operating the biggest public LoRaWAN network as well as dozens of private networks in all colors and flavors. We don't trade quality for a shorter time-to-market. We don't take technical shortcuts. It's quality first, built for the future, and that will pay off.
A distinctive feature of V3 is its flexible deployment models. The same codebase, the same API, scalable from a small on-premises deployment to a multi-region, multi-tenant, multi-instance, supported, SLA backed and hosted deployment.
- Hosted is our flagship V3 SaaS offering; fully featured, hosted and SLA backed
- Private Cloud is the V3 Stack in your own AWS, Azure or GCP account, with integrations with their IoT platforms
- Images and Binaries allow for deploying the V3 Stack on-premises, on various architectures, enabling single binary deployments on gateways to custom Kubernetes deployments in the cloud
- Open Source allows you to compile from source and run the Stack on your development machine
- Community Network is the free to use The Things Network public community network
*not yet in MVP.
The V3 architecture follows exactly the LoRaWAN Network Reference Model. Not only in the naming of the components, but also in their responsibilities and future LoRaWAN Back-end Interfaces support. This allows for working with external Join Servers and Network Servers for full control over security.
Private MVP Milestone
Our first milestone is Private MVP, scheduled for this quarter; the minimum viable product for private networks. The V3 MVP is intended to be used in new deployments and adapting existing applications. Also, we'll provide tooling to run V2 and V3 side-by-side, using the same gateway infrastructure.
We're close to releasing the MVP. Key things that still need to be done:
- Finalizing the Console. Interfacing in the MVP is primarily API and command line
- Integration testing
- Finalizing Network Server MAC layer, especially downlink scheduling
- Finalizing validation of API calls
- Bridging to V2 and V3 for side-by-side deployments
- Making everything GitHub ready; housekeeping and documentation
The Private MVP is just the first milestone. V3 is ongoing development, with short iterations. Based on our roadmap and your feedback, we make things ready for production readiness and the other deployment models.
V3 is API first. You will be able to do everything through the API, allowing full integration with your solution.
I already want to share the V3 API with you. This helps understanding how your solution can interact with V3, and gives an idea of how powerful V3 is.
You can request the Markdown documentation and Swagger file. We may be adding or changes a few things the coming weeks. With the MVP release, this API is frozen and we won't be introducing breaking changes.
The V3 architecture has a nice separation of components: the Identity Server (IS), Gateway Server (GS), Join Server (JS), Network Server (NS) and Application Server (AS), with their own API. The IS, JS, NS and AS each have a device registry API, as they all manage parts of the end device. For example, you'll use the JS API for managing keys, while you'll use the AS API for payload formatters. Our SDKs, part of a later milestone, will make it easy to work with such a distributed deployment.
The next update is in two weeks!
Please invite colleagues and partners to this list to stay updated.
CTO and Co-founder, The Things Industries