Loading...

Performability Modeling and Analysis in Grid Computing

Entezari Maleki, Reza | 2015

752 Viewed
  1. Type of Document: Ph.D. Dissertation
  2. Language: Farsi
  3. Document No: 46971 (19)
  4. University: Sharif University of Technology
  5. Department: Computer Engineering
  6. Advisor(s): Movaghar Rahimabadi, Ali
  7. Abstract:
  8. In this thesis, three different mathematical models named Markov Reward Model (MRM), Stochastic Reward Net (SRN) and Stochastic Activity Network (SAN) are used to model and evaluate the performability of grid computing environments consisting of many grid resources. The proposed models consider the arriving and servicing process of grid tasks inside a resource together with the failure-repair behavior of processors of the resource. Since the proposed MRM cannot be extended to model a grid environment with some realistic assumptions, we switch to use SRNs in modeling a single grid resource with more number of processors. Although the proposed SRN models for a single grid resource can appropriately model and evaluate the combined performance and dependability of a resource, they encounter state space explosion problem whenever they get together to capture a real grid environment. To solve this problem, two approximate models are proposed using folding and fixed-point iteration methods to appropriately estimate the results of the exact model of a grid environment. The results obtained from numerical analysis of the proposed models and simulating the corresponding systems show that the approximate models can estimate the exact model properly. Afterwards, two scheduling approaches are presented to use the results gained from the previous step to schedule grid applications to grid resources. The first approach considers independent grid programs and uses curve fitting to find the service time of each resource to the set of independent programs, where the second approach applies the u-function technique to compute the probability mass function of service time of entire grid to a workflow application consisting of many dependent programs. Finally, both approaches apply heuristic algorithms to solve the scheduling problem which is an NP-complete problem
  9. Keywords:
  10. Efficiency ; Dependability ; Computational Grids ; Stochastic Activity Networks ; Task Scheduling Algorithm ; Stochastic Reward Nets (SRN)

 Digital Object List

 Bookmark

No TOC