IDS 200 Lecture Notes - Lecture 26: Service-Level Agreement, In Essence, Network Congestion
Document Summary
To help smooth out problems with latency and jitter, video streaming applications use buffering. In essence, the application builds up a reserve of data, typically several seconds. Latency is handled until the latency extends past the buffer"s end. Jitter is handled until the sum of jitter & latency extends past the buffer"s end. Lost packets might be recovered if resent & received by the buffer"s end. Since quality of service inherently means different things to different users of different applications: It"s not a monolithic variable, although for simplicity it might be represented as one. It"s more correct to present qos as a set of metrics including latency, jitter, packet throughput, etc. Alternately, qos can be considered in terms of: How well performance meets a desired design or operating standard (like a service level agreement) Many networks experience significant but transient bursts of heavy traffic. Can cause corresponding transient drops in qos due to network congestion.