In a classical deployment, a service can be implemented as several instances of the same application (e.g., the same code) all running in parallel. Some specific applications, such as for connected vehicles, can involve several instances of the same application running on well-defined compute nodes that are physically located at different locations. In some instances, it may be desirable (e.g., for performance reasons) for a connected vehicle to connect to a nearby application instance (i.e., having a smallest round trip time (RTT)). This typically involves a user disconnecting from a current instance, finding or being redirected to another instance, establishing a connection with the new instance, etc. In classical approaches, such operations involve a control plane providing instructions to a user (e.g., by sending the user an identifier of a new instance) thereby triggering the process. However, techniques herein provide for the ability to simplify this problem by giving full control to a platform, in an almost fully transparent manner to the device, which provides for the ability to dynamically route a moving device to an appropriate server in a fully distributed manner.

Creative Commons License

Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.