Skip to content

Architecture Patterns#

The RTA API Specification does not require any specific architecture or technology stack — but there are some common patterns.

This page illustrates some architectures supported by services from our implementation toolkit.

Legend

Development#

RTA Server#

RTA Server provides a simple, single-process deployment to experiment to explore the RTA API Specification and gain familiarity with Sessions, Configuration and the data types.

Microservices#

The Toolkit Services are designed be to run as microservices within a container orchestrator, like Kubernetes.

Here is the same capability, split into individual components:

All these services are horizontally-scalable, facilitating high-availability deployments.

OAuth 2.0 Security#

ATLAS and the Toolkit Services can use OAuth 2.0 with multi-factor authentication:

The Gateway Service proxies the authorization bearer token through to the other microservices.

Data Adapter Service#

RTA does not require data format-shifting.

Instead of copying the data with a loader, you can index it in place and deploy a Data Adapter Service to read the store:

McLaren provide an InfluxDB Data Service in the implementation toolkit:

Custom file formats can be served using exactly the same pattern:

Hetrogeneous formats can all be handled in one deployment:

Continuous Ingest and Live Streaming#

The diagrams above show data being indexed after it has arrived. But this can be integrated into existing ingest pipelines.

Calls to the Toolkit Services are made using gRPC, which is supported in a wide range of languages — so it is easy to publish configuration and session metadata from an existing framework.

This is not limited to metadata. You can live-stream data to ATLAS:

A copy of the data is streamed across to Redis, and connected to clients via Stream Service web sockets:

This streaming strategy can provide a fast bypass when the store has high latency — for example, file storage:

Remote Access#

The web services are specifically designed for high-latency operation across the WAN or Internet.

Connections are secured with TLS and OAuth 2.0, so users can connect to a data center or cloud deployment without a VPN:

Many OAuth 2.0 providers offer identity federation, so subsets of data can be shared securely across company boundaries: