Abstract
A distributed group-pricing algorithm is presented for market-based resource scheduling in grid context, which is integrated with quick convergence of the centralized simultaneous pricing algorithm and scalability of the distributed WALRAS algorithm. Firstly, an agent-based framework for resources pricing is described: resource agent and auctioneer for groups of resources. Secondly, the distributed group-pricing algorithm is introduced: resources in the system are grouped according to the degree of interaction, and each auctioneer adjusts the price of one group of resources respectively until the excess demand of resources becomes zero when the demand and supply of resources of the system changes. Finally, the performance of the WALRAS algorithm and the presented distributed group-pricing algorithm is compared by experiments, and the experimental results indicate that the distributed group-pricing algorithm can attain better performance than the WALRAS algorithm.
| Original language | English |
|---|---|
| Pages (from-to) | 1151-1156 |
| Number of pages | 6 |
| Journal | Jisuanji Yanjiu yu Fazhan/Computer Research and Development |
| Volume | 41 |
| Issue number | 7 |
| State | Published - Jul 2004 |
| Externally published | Yes |
Keywords
- Distributed pricing
- Grid computing
- Pricing algorithm