| D. Feitelson and M. Naaman. Self-tuning systems. IEEE Software, 16(2):52--60, 1999. |
....jobs are in the system. Depending on that, IVS switches between FFDH (more batch jobs) or FFIH (more interactive jobs) If the system is not saturated, FCFS is used. IVS was never implemented and tested in a real environment, as the project finished with the Ph.D. thesis. Feitelson and Naaman [3] published work about self tuning systems in 1999. Modern operating systems are highly parameterized, so that the administrative sta# is forced to use a trial and error approach for optimizing these parameters. A better way would be to automate this process. The idea is to look at past information ....
....The Self Tuning dynP Scheduler One problem still is, that a long lasting trial and error process is needed, to find proper values for the two bounds. The fact that our simulation environment is now working with full schedules inspired by Feitelsons and Naamans work about self tuning systems [3], brought us to the idea of the self tuning dynP scheduler: Let the scheduler generate full schedules for each of the three strategies in every scheduling step. And switch to that policy which generates the best schedule for the current situation. core self tuning dynP algorithm( mySchedule ....
D. G. Feitelson and M. Naaman. Self-Tuning Systems. In IEEE Software 16(2), pages 52--60, April/Mai 1999.
....of each decision step the scheduler computes the full schedule for each of the three policies and then rates each schedule according to a quality parameter. Then the scheduler changes to that policy with achieved the best performance in this step. Similar work was done by Feitelson and Naaman [2]. They used genetic algorithms and a fitness function to search for optimal parameter values for their scheduling algorithms. So with that the scheduler can automatically choose the best sorting policy without any parameter input. The core algorithm (decider) of the self tuning dynP scheduler ....
D. G. Feitelson and M. Naaman. Self-Tuning Systems. In IEEE Software 16(2), pages 52--60, April/Mai 1999.
....good as with LJF for both job sets. As already mentioned the performance of dynP depends on the proper settings of the bounds. To ease the usability of the dynP scheduler we are currently working on an adaptive version of dynP with does not require anymore startup parameters. Self tuning systems [5] might be used for this. In the future we will also add the dynP scheduler to CCS and use it for everyday work. This will show, if the simulated results from this paper can also be achieved in a real environment. ....
D.G. Feitelson and M. Naaman. Self-Tuning Systems. In IEEE Software 16(2), pages 52--60, April/Mai 1999.
No context found.
D. G. Feitelson and M. Naaman, "Self-tuning systems ". IEEE Softw. 16(2), pp. 52--60, Mar/Apr 1999.
No context found.
D. Feitelson and M. Naaman. Self-tuning systems. IEEE Software, 16(2):52--60, 1999.
No context found.
D. G. Feitelson and M. Naaman. Self-Tuning Systems. In IEEE Software 16(2), pages 52--60, April/Mai 1999.
No context found.
D. G. Feitelson and M. Naaman. Self-Tuning Systems. In IEEE Software 16(2), pages 52--60, April/Mai 1999.
No context found.
Feitelson, D. G. and Naaman,, M.: Self-tuning Systems. IEEE Software. (1999) 52-60
No context found.
D. G. Feitelson and M. Naaman. Self-Tuning Systems. In IEEE Software 16(2), pages 52--60, April/Mai 1999.
No context found.
D. G. Feitelson and M. Naaman. Self-Tuning Systems. In IEEE Software 16(2), pages 52--60, April/Mai 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