{"title":"5的结论","authors":"R. Dssouli","doi":"10.7788/9783412506865-007","DOIUrl":null,"url":null,"abstract":"26 tion. DRS allows to recover from delay violations of specific components for each individual data unit which is transmitted. The role of DRS is to avoid, if possible, end-to-end delay violation, but it introduces an important overhead. The proposed schemes, especially CRS, are able to support a graceful degradation. Indeed, when the QoS manager fails to find an alternate configuration which supports the initially agreed QoS, it may select a configuration which supports a lower QoS in a graceful way, e.g., continue playing black&white video instead of the initial color video. A scheme which supports the characteristics of RDS, RRS and CRS, may be more efficient. As an example, for the delay parameter, the following scenario may be used: (1) DRS is used to maintain end-to-end delay of each data unit. (2) If a component is in difficulty to supports its delay requirements, RRS is used to reconfig-ure the resource distribution of the components in a way to satisfy end-to-end delay requirements. The delay considered is the delay averaged over some measurement interval. This operation is executed in parallel with operation (1). (3) If the operation (2) does not succeed, CRS is used to find an alternate configuration able to support end-to-end requirements. Then the QoS manager performs a transition from the current configuration to the new one. Currently we are implementing a simple instantiation of CRS to study its feasibility and its performance in a real environment taking remote access to news-on-demand as a case study [Haf 96]. To detect QoS violations, we make use of a monitoring tool [Som 96] that we have implemented at the transport level. Our experimental platform consists of a network employing point-to-point links coupled to a high speed ATM switch (Newbrige) to form an ATM LAN and two IBM RS/6000, running AIX. An Ethernet is also used to connect the machines, Acknowledgment We would like to thank R.Velthuys from IBM Toronto, and A.Khoumsi from University of Montreal for fruitful discussions on a draft of this paper. 25-The QoS agent processes the data unit, e.g., a decoder decodes the data unit and a network transports the data unit, while doing its best (depending on its available resources) to satisfy a delay of =-ViolationDegree; else (ViolationDegree)-The QoS agent processes the data unit; in this case the QoS agent may relax its commitment by ViolationDegree (or fraction of ViolationDegree), that is, it has only to …","PeriodicalId":129866,"journal":{"name":"Risk-Sharing Finance","volume":"2 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2020-11-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"5 Conclusion\",\"authors\":\"R. Dssouli\",\"doi\":\"10.7788/9783412506865-007\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"26 tion. DRS allows to recover from delay violations of specific components for each individual data unit which is transmitted. The role of DRS is to avoid, if possible, end-to-end delay violation, but it introduces an important overhead. The proposed schemes, especially CRS, are able to support a graceful degradation. Indeed, when the QoS manager fails to find an alternate configuration which supports the initially agreed QoS, it may select a configuration which supports a lower QoS in a graceful way, e.g., continue playing black&white video instead of the initial color video. A scheme which supports the characteristics of RDS, RRS and CRS, may be more efficient. As an example, for the delay parameter, the following scenario may be used: (1) DRS is used to maintain end-to-end delay of each data unit. (2) If a component is in difficulty to supports its delay requirements, RRS is used to reconfig-ure the resource distribution of the components in a way to satisfy end-to-end delay requirements. The delay considered is the delay averaged over some measurement interval. This operation is executed in parallel with operation (1). (3) If the operation (2) does not succeed, CRS is used to find an alternate configuration able to support end-to-end requirements. Then the QoS manager performs a transition from the current configuration to the new one. Currently we are implementing a simple instantiation of CRS to study its feasibility and its performance in a real environment taking remote access to news-on-demand as a case study [Haf 96]. To detect QoS violations, we make use of a monitoring tool [Som 96] that we have implemented at the transport level. Our experimental platform consists of a network employing point-to-point links coupled to a high speed ATM switch (Newbrige) to form an ATM LAN and two IBM RS/6000, running AIX. An Ethernet is also used to connect the machines, Acknowledgment We would like to thank R.Velthuys from IBM Toronto, and A.Khoumsi from University of Montreal for fruitful discussions on a draft of this paper. 25-The QoS agent processes the data unit, e.g., a decoder decodes the data unit and a network transports the data unit, while doing its best (depending on its available resources) to satisfy a delay of =-ViolationDegree; else (ViolationDegree)-The QoS agent processes the data unit; in this case the QoS agent may relax its commitment by ViolationDegree (or fraction of ViolationDegree), that is, it has only to …\",\"PeriodicalId\":129866,\"journal\":{\"name\":\"Risk-Sharing Finance\",\"volume\":\"2 1\",\"pages\":\"0\"},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2020-11-04\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Risk-Sharing Finance\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/10.7788/9783412506865-007\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Risk-Sharing Finance","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.7788/9783412506865-007","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
26 tion. DRS allows to recover from delay violations of specific components for each individual data unit which is transmitted. The role of DRS is to avoid, if possible, end-to-end delay violation, but it introduces an important overhead. The proposed schemes, especially CRS, are able to support a graceful degradation. Indeed, when the QoS manager fails to find an alternate configuration which supports the initially agreed QoS, it may select a configuration which supports a lower QoS in a graceful way, e.g., continue playing black&white video instead of the initial color video. A scheme which supports the characteristics of RDS, RRS and CRS, may be more efficient. As an example, for the delay parameter, the following scenario may be used: (1) DRS is used to maintain end-to-end delay of each data unit. (2) If a component is in difficulty to supports its delay requirements, RRS is used to reconfig-ure the resource distribution of the components in a way to satisfy end-to-end delay requirements. The delay considered is the delay averaged over some measurement interval. This operation is executed in parallel with operation (1). (3) If the operation (2) does not succeed, CRS is used to find an alternate configuration able to support end-to-end requirements. Then the QoS manager performs a transition from the current configuration to the new one. Currently we are implementing a simple instantiation of CRS to study its feasibility and its performance in a real environment taking remote access to news-on-demand as a case study [Haf 96]. To detect QoS violations, we make use of a monitoring tool [Som 96] that we have implemented at the transport level. Our experimental platform consists of a network employing point-to-point links coupled to a high speed ATM switch (Newbrige) to form an ATM LAN and two IBM RS/6000, running AIX. An Ethernet is also used to connect the machines, Acknowledgment We would like to thank R.Velthuys from IBM Toronto, and A.Khoumsi from University of Montreal for fruitful discussions on a draft of this paper. 25-The QoS agent processes the data unit, e.g., a decoder decodes the data unit and a network transports the data unit, while doing its best (depending on its available resources) to satisfy a delay of =-ViolationDegree; else (ViolationDegree)-The QoS agent processes the data unit; in this case the QoS agent may relax its commitment by ViolationDegree (or fraction of ViolationDegree), that is, it has only to …