Dual mode operation is common in computing. Take edge routers for instance. As opposed to core routers, Edge routers additionally perform admission control. To provide more context to routers and admission control, lets' review quality of service.
Internet has been serving a single class of service called the Best Effort service. There is no assurances about delivery. Such assurances based on rate and throughput are defined as quality of service. In the absence of quality of service (QoS), applications have to tolerate losses and delays and adapt to congestion. Should we modify the applications to be more adaptive or should we modify the internet to support QoS.
Applications that adapt to the performance of network are called "tolerant real-time" and those that demand hard limits on performance are called "intolerant real-time".
Quality of Service guarantees are based on sharing and congestion studies. In order to study congestion, packets are marked for router to distinguish between different classes and new router policies to treat packets differently. QoS guarantees provide isolation for one class from other classes When allocating different bandwidths to each application flow, bandwidth may not be efficiently used. Therefore another tenet of Qos guarantees is that while providing isolation to different classes, it is preferred to utilize the resources as efficiently as possible. Since most networks have a link capacity that cannot be exceeded. there is a need to control admission. Applications can declare their requirements, and network can block if it cannot provide the resources.
Therefore Quality of Service are provided using packet classification, Isolation: scheduling and policy, high resource utilization and call admission
IntServ and DiffServ are models of providing quality of Service. IntServ stands for Integrated services and DiffServ stands for Differentiated Services. In the IntServ model, QoS is applied on a per flow basis and addresses concerns such as business model and charging. Even in mobile phone networks, this is evident when certain billing options are desirable but not possible. In Differentiated services, the emphasis is on scalability, flexible service model, and simpler signaling. Scalability here means we don't track resources for each of the possibly large number of flows. Service model means we provide bands of service such as platinum, gold and silver.
Courtesy: lecture notes CMU
Internet has been serving a single class of service called the Best Effort service. There is no assurances about delivery. Such assurances based on rate and throughput are defined as quality of service. In the absence of quality of service (QoS), applications have to tolerate losses and delays and adapt to congestion. Should we modify the applications to be more adaptive or should we modify the internet to support QoS.
Applications that adapt to the performance of network are called "tolerant real-time" and those that demand hard limits on performance are called "intolerant real-time".
Quality of Service guarantees are based on sharing and congestion studies. In order to study congestion, packets are marked for router to distinguish between different classes and new router policies to treat packets differently. QoS guarantees provide isolation for one class from other classes When allocating different bandwidths to each application flow, bandwidth may not be efficiently used. Therefore another tenet of Qos guarantees is that while providing isolation to different classes, it is preferred to utilize the resources as efficiently as possible. Since most networks have a link capacity that cannot be exceeded. there is a need to control admission. Applications can declare their requirements, and network can block if it cannot provide the resources.
Therefore Quality of Service are provided using packet classification, Isolation: scheduling and policy, high resource utilization and call admission
IntServ and DiffServ are models of providing quality of Service. IntServ stands for Integrated services and DiffServ stands for Differentiated Services. In the IntServ model, QoS is applied on a per flow basis and addresses concerns such as business model and charging. Even in mobile phone networks, this is evident when certain billing options are desirable but not possible. In Differentiated services, the emphasis is on scalability, flexible service model, and simpler signaling. Scalability here means we don't track resources for each of the possibly large number of flows. Service model means we provide bands of service such as platinum, gold and silver.
Courtesy: lecture notes CMU
No comments:
Post a Comment