T. Abdelzaher , "QOS Adaptation in Real-Time Systems", PhD Thesis, University of Michigan, 1999.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....on a node. Both the schedulers collaborate to adapt to changing system and network conditions by runtime monitoring. The middleware supports standard real time scheduling algorithms, such as rate monotonic and earliest deadline first algorithms, as well as priority based scheduling. Adaptware [1] is another QoS adaptive middleware that uses a feedback based control theoretic approach to schedule resources. Clients specify a range of acceptable QoS through QoS contracts with the server. Adaptware uses those contracts to adapt the QoS delivered to a client, so as to maximize the aggregate ....

T. Abdelzaher. QoS Adaptation in Real-Time Systems. PhD thesis, University of Michigan, Ann Arbor, August 1999.

....files (i.e. if dynamic content is statistically insignificant or served from a separate server) one can express server utilization U as a function of the served request rate R and delivered byte bandwidth W . Both of these variables can be measured very accurately in a computing server. In [2], we showed that bottleneck resource utilization is given by: U = aR bW (1) where a and b are constants which can be computed by a linear regression. The intuition behind Eq. 1) comes from the fact that the resource requirements C i of serving a file are composed from a fixed overhead plus a ....

T. F. Abdelzaher. QoS-Adaptation in Real-Time Systems. PhD thesis, University of Michigan, Ann Arbor, Michigan, August 1999.

....U [l] to the least squares estimator which implements the set of equations 3.11. Utilization, in our implementation, is measured in an OS independent manner by computing the fraction of time the server is busy serving requests. An algorithm for measuring utilization in this manner is described in [2]. If there is more that one resource of interest to the server, each resource has its own estimator. The utilization U j [l] must be measured for each resource j independently, and the values m l and U j [l] are fed to the respective estimator. The jth estimator refines the estimate of parameters ....

T. F. Abdelzaher. QoS-Adaptation in Real-Time Systems. PhD thesis, University of Michigan, Ann Arbor, Michigan, August 1999.

....files (i.e. if dynamic content is statistically insignificant or served from a separate server) one can express server utilization U as a function of the served request rate R and delivered byte bandwidth W . Both of these variables can be measured very accurately in a computing server. In [2], we showed that bottleneck resource utilization is given by: U = aR bW (1) where a and b are constants which can be computed by a linear regression. The intuition behind Eq. 1) comes from the fact that the resource requirements C i of serving a file are composed from a fixed overhead plus a ....

T. F. Abdelzaher. QoS-Adaptation in Real-Time Systems. PhD thesis, University of Michigan, Ann Arbor, Michigan, August 1999.

....measurements. In the common case where the web server serves mostly static files (web pages) one can express server utilization U as a function of the served request rate R and delivered byte bandwidth W . Both of these variables can be measured very accurately in a computing server. In [2], we prove that: U = aR bW (1) where a and b are constants which can be computed by a linear regression. The intuition behind Eq. 1) comes from the fact that the resource requirements C i of serving a file are composed from a fixed overhead plus a variable overhead that depends on the length x ....

T. F. Abdelzaher. QoS-Adaptation in Real-Time Systems. PhD thesis, University of Michigan, Ann Arbor, Michigan, August 1999.

No context found.

T. Abdelzaher , "QOS Adaptation in Real-Time Systems", PhD Thesis, University of Michigan, 1999.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC