By Sebnem Ozer, Ph.D., Comcast
The requirements of emerging interactive real-time services and changes in online usage patterns impose entirely different network challenges that Internet Service Providers need to overcome. Cable networks are going through a big transition to the next-generation 10G technologies with substantial speed increases, that can meet the online traffic volumes created by these services accumulatively. However, the new quality of experience judge will not praise or condemn the network operators only by their speed but also by their consistent support of low latency. Therefore, 10G technologies need to address a fundamental redesign of traffic classification and latency monitoring, prediction and optimization.
Unprepared Internet Service Providers (ISPs) that design their architecture for mean and median values instead of peaks cannot support the interactivity of real-time services and cannot avoid the impact of these huge data volumes on other services.
In this paper, we will discuss the low latency services that are still evolving today, such as cloud gaming, video/voice conferencing and live video streaming, as well as emerging applications with progressively more interwoven human and machine interactions. We will first cover current network features and tools that can be used to measure, monitor and manage the latency of today’s networks. We will then describe the next steps to support new Low Latency (LL) services by applying D3.1 features and a LL service differentiation framework. To support the LLD (Low Latency DOCSIS) features, traffic classification and monitoring must be redesigned and inherent rules may need to be replaced. Lastly, we will provide guidelines to deliver low latency services with the most efficient and future-proof investments.