The RackN Digital Rebar platform helps organizations build end-to-end infrastructure as code (IaC) workflows for any infrastructure type. Running IaC is critical to rapidly deploying infrastructure changes. It can prevent long infrastructure delays that impact the software development team’s productivity. But what happens when someone changes the code in one site when you are managing many sites? Automating one site at a time is not good enough, so how can you keep hundreds of automated sites synchronized?
Digital Rebar Federated Management
When we started the fourth Digital Rebar generation (v4.x), we recognized that operators needed something different to synchronize automated sites than the common star-and-spoke control plane. They could not afford to have distributed and independent sites dependent on a remote location for operations. Even worse, our customers were scaling such that a centralized solution would quickly fail.
To respond, we created the Digital Rebar federated management architecture. In the product, we call it Multi-Site Manager. With this design, each Digital Rebar site can be completely autonomous. At the same time, remote managers are able to create aggregated views of multiple sites.
How does Federated Management work?
Our unique IaC content system makes it practical for autonomous sites to share automation without needing to be tightly coupled. But the most remarkable feature of the design is our integration API proxy and forward. This allows a “single pane of glass” manager API without having two APIs or compromising any site’s autonomy.
In other words, you can make the exact same API calls against a manager that you’d make against a connected site. How did we do that?
- First, the API proxy feature allows the managers to maintain read-only mirrors of other site’s endpoints.
- All API interactions, including the response, are exactly the same regardless of which endpoint receives the call.
- The manager site proxies the request to the site that controls the infrastructure and returns the API response.
- Finally, the manager site gets updates to its mirrored copy of the state.
- The design also allows you to have multiple managers. This eliminates the bottleneck “star” manager in traditional star-and-spoke designs.
This video gives more insight into the Multi-site Manager.
This is officially patented technology!
We’re proud to announce that the US Patent Office has recognized this innovative architecture with a patent award. The feature has been in Digital Rebar for over four years, even though the patent was just granted. That means that operators have been using and scaling on this feature since version 4.5.
The full patent name is “Distributed federation of endpoints with proxy synchronization”. You can read more of how federated management (aka Multi-Site Manager) is implemented in the documentation.
If your organization is building out an IaC practice, how are you keeping those of automated sites synchronized once they are built? Our patented Multi-Site Manager may be the solution you need. We’d love to show you more! Click here to schedule a demo with us.
ABSTRACT
An endpoint of a distributed federation with proxy synchronization including a data center infrastructure, a storage, and an endpoint. The storage stores a state of the data infrastructure and further stores a mirrored state for each of at least one other endpoint. The endpoint includes a communication interface for communicating via a communication network, where the endpoint, in response to receiving a command via the communication interface for changing the mirrored state, forwards the command towards an endpoint that owns the mirrored state via the communication interface. Commands may be forwarded directly or indirectly via one or more intermediary endpoints. An owner endpoint receives a command, updates its local state, and sends one or more events to one or more proxy endpoints to update corresponding mirrored states. A restricted proxy endpoint may store a partial mirrored state. The federation may support bidirectional sharing, synchronization, and resource data sharing. 
