The Research On Network QoS Monitor Module And Degradation Locating

In recent years, Internet rapidly developed and a variety of applications in the Internet network, particularly with regard to Multimedia, VoIP and online video-on-demand applications, are greatly increasing. In particular, some video-based websites such as youtube, youku and tudou are so popular that there is a trend that these video-websites will replace traditional websites and become the mainstream media websites in everyday life. The video based network is a large scale, widely distributed network. Actually, current Internet only provides a best-effort delivery system (Best-effort delivery system). The QoS degradations phenomenon may occur during multimedia transmission, which could not satisfy SLA (Service Level Agreement) requirements. In other words, traditional network will not have a mechanism to ensure its QoS, while the requirement of guaranteeing QoS is becoming more and more increasing. Both content providers and network service providers are trying their best to establish a mechanism for monitoring quality of service and locating degradations, whose goal is to ensure their customers'quality of service and improve their service level. In order to guarantee QoS requirement, we have to monitor real-time QoS parameters. The purpose of monitoring QoS is to locate these QoS degradation nodes and provide network manager with necessary QoS degradation information, which will help network manager adjust resource allocation, meeting SLA requirement.In this paper, we introduce current QoS monitor models and methods of degradation locating. It includes end to end QoS monitoring model and distributed QoS monitoring model. When it comes to current methods of degradation locating, we introduce a end to end degradation locating method based on Ping and Tracert, a distributed degradation locating method based on key network segment, an automotive degradation locating method based on rule and an intelligent degradation locating method based on case reasoning. In the aspect of QoS monitoring model: the end to end QoS monitoring model is very convenient, flexible and effective. It does not need to concern about middle way nodes and just monitors an application flow's source node's and destination node's QoS level. But this method could not provide us more information about middle way nodes'QoS conditions, so we could not locate which middle way nodes'QoS level degraded; The distributed QoS monitoring model based on SNMP(Simple Network Management Protocol) provides us a very standard method for real-time monitoring and managing network. Although it is comprehensive, flexible and easy to maintain, the central manager workstation is its bottleneck. Besides, this method will bring large amount of additional costs when monitoring QoS; The monitor model based on relevant monitor provides us a mechanism for finding relevant QoS monitors regard to a specific application flow. Actually, we adopt this mechanism and regard it as a very effective tool in our proposed distributed QoS monitor model. In the aspect of degradation locating, the obvious benefit of Ping and Tracert degradation locating method is convenient and fast. But it is just available when trying to test whether network is accessible or not. It could not meet our requirement of degradation locating in QoS level. Actually it is an effective degradation locating method in physical and data link layer; The degradation method based on key network segment is trying to find the first QoS degradation node in a specific application flow. It is very accurate, but has low efficiency and needs recursive running; The degradation locating method based on rule is good at recognizing degradation nodes and filtering unnecessary degradation information. Besides, it has a certain automotive ability. But the rate of rule patterning is low and rule base is so large that it is hard to maintain; The degradation locating method based on case reasoning tries to abstract the essence of a specific case and add the successful solution to this kind of degradation problem into case base. The scale of its case base is not large and easy to maintain. But sometimes, it is hard to find a reliable abstract model.After studying and analyzing previous QoS monitoring models, we propose an integration distributed QoS monitoring model. It imports SNMP central monitor system and mobile agent monitor technology. The main purpose of this QoS monitor model is to decrease monitoring cost, response time, solve central bottleneck and find QoS monitor location according to a specific application flow. In addition, we study previous QoS degradation locating methods and find there could be one or more degradations nodes in a specific application flow. If current node's and its previous node's QoS level could not meet SLA requirement, we could not say that the current node's QoS level decreases. Because, QoS degradation has the characteristics of correlation and transmission, e.g., jitter and delay. Therefore, it is entirely possible that the previous QoS degradation node causes the current node's QoS level could not meet SLA requirement. In this situation, if we use certain mechanisms to adjust the resource allocation for improving previous QoS degradation node's QoS level, the phenomenon of the current node's QoS degradation could disappear. How to describe this kind of correlation and locate those nodes, which actually cause QoS degradations are the main problem we need to concern in this paper. According to the above problem, we propose a multi nodes degradation locating method based on correlation compensating. It depends on the correlation among different QoS parameters, e.g., bandwidth, delay, jitter and loss package rate, using an adaptive compensating algorithm to dilute this kind of correlation in order to find degradation.In the simulation experiment part, we compare our proposed integration distributed QoS monitoring model with previous distributed QoS monitor models in the aspects of data cost and response time. In addition, we compare our MDCC with previous KNS degradation locating method in the aspects of accuracy and effectiveness. As a result illustrated by our simulation, our proposed integration distributed QoS monitor model and MDCC degradation locating method have better efficiency compared with previous QoS monitor models and degradation locating methods.