Evolved Cloud Collaboration
We looked into how the current centralized architecture of video conferencing platforms could be re-designed to improve quality and optimize the use of resources by adopting a distributed architecture.
Have you ever been in a video conference where the flow of the discussion was interrupted by annoying quality issues with audio and video? We believe that the next step in the evolution of the cloud and the architecture of video conferencing platforms could reduce the likelihood of such issues occurring and thus improve user experience.
The problem of low or varying quality of an application such as video conferencing that has users connecting to it from various parts of the world is familiar to many users. These problems are becoming even more common today when video services are moving to the cloud and start to be offered as a service.
At Ericsson Research, we looked into how the current centralized architecture of video conferencing platforms could be re-designed to improve quality and optimize the use of resources by adopting a distributed architecture. We also looked into services that a distributed cloud platform can offer to applications such as video conferencing to better facilitate their run-time adaptation.
We moved away from the traditional centralized design where all the components of the video platform are running in one location to a distributed architecture where the components of the platform are distributed to many locations as close to the users as possible, that is, to geographically distributed data centers.
In our solution, an intermediate layer between the application and the cloud infrastructure makes it possible for the application to interact with the cloud platform at run-time and adjust the required resources according to the needs of the application and the conditions of the environment.
In conferences that have geographically distributed users, and where the media servers are running in one centralized location, the operator experiences high bandwidth usage on long-haul links and the end users experience high latencies. In contrast, in a distributed cloud environment, media services can be provided close to where the users are located, which results in lower bandwidth usage, lower latencies, and better experience for the users. Such a distributed architecture also provides robustness by removing any single point of failure.
Traditionally, media servers providing enterprise communication services are deployed on the customer premise. In this CAPEX-based model, the enterprise owns the infrastructure needed for the communication service. Currently, there is a growing interest towards an OPEX-based model where communication services are delivered to the enterprise from the cloud using the “as-a-service” delivery model. Such as-a-service delivery is further optimized by a distributed cloud platform and a distributed application architecture.
Further, in an on-premise deployment, there is often a need to over-provision the media servers so that peak load can be handled. On-premise deployment also results in a need for manual configuration and maintenance. The benefits that a cloud-based virtualized video platform that interacts with the cloud brings include close to unlimited scalability from small to large workloads, and high amount of automation.
Our cloud and video platforms interact with each other in order to achieve advanced features such as initial optimal placement of virtual machines, SLA (Service Level Agreement) driven application deployment, distributed image handling, auto-scaling, a triggering mechanism to react to changes in the environment at runtime, and location service to inform the application about the most optimal location for hosting its components. Our distributed cloud orchestration solution prototype is capable of managing resources in geographically distributed data centers and satisfies the stringent needs of any real-time application.
This will enable cloud-based video conferencing services that provide high user-experience and minimize the usage of long-haul bandwidth in conferences having geographically distributed participants, reducing the load on operators’ networks.
We demonstrated this in Barcelona, at the MWC and here is the Slideshare presentation from this demonstration.
Azimeh Sefidcon, Jouni Mäenpää from Ericsson Research