Competitive distributed file allocation
STOC '93 Proceedings of the twenty-fifth annual ACM symposium on Theory of computing
Randomized routing and sorting on fixed-connection networks
Journal of Algorithms
On the benefit of supporting virtual channels in wormhole routers
Proceedings of the eighth annual ACM symposium on Parallel algorithms and architectures
Universal continuous routing strategies
Proceedings of the eighth annual ACM symposium on Parallel algorithms and architectures
Universal algorithms for store-and-forward and wormhole routing
STOC '96 Proceedings of the twenty-eighth annual ACM symposium on Theory of computing
Universal O(congestion + dilation + log1+&egr;N) local control packet switching algorithms
STOC '97 Proceedings of the twenty-ninth annual ACM symposium on Theory of computing
Distributed paging for general networks
Journal of Algorithms
Data management in networks: experimental evaluation of a provably good strategy
Proceedings of the eleventh annual ACM symposium on Parallel algorithms and architectures
Caching in networks (extended abstract)
SODA '00 Proceedings of the eleventh annual ACM-SIAM symposium on Discrete algorithms
Computers and Intractability; A Guide to the Theory of NP-Completeness
Computers and Intractability; A Guide to the Theory of NP-Completeness
Provably Good and Practical Strategies for Non-Uniform Data Management in Networks
ESA '99 Proceedings of the 7th Annual European Symposium on Algorithms
Exploiting Locality for Data Management in Systems of Limited Bandwidth
FOCS '97 Proceedings of the 38th Annual Symposium on Foundations of Computer Science
The Scalable Coherent Interface (SCI)
IEEE Communications Magazine
Inferring tree topologies using flow tests
SODA '03 Proceedings of the fourteenth annual ACM-SIAM symposium on Discrete algorithms
Hi-index | 0.00 |
A hierarchical bus network T = (V, E) uses hierarchically, tree-like connected buses as a communication network. New communication technologies like SCI (Scalable Coherent Interface) (see, e.g., [6, 7]) make such networks very attractive, because they allow their easy construction and guarantee reasonable communication performance. Such networks can be modeled as tree networks: leaves correspond to processors, inner nodes to buses, edges to switches, and bandwidths of inner nodes and edges are related to bandwidths of buses and switches, respectively.In this paper we address the problem of static data management. Given a set of shared data objects X and the read and write frequencies from the processors to the shared data objects, the goal is to compute a (maybe redundant) placement of the shared data objects to the processors, such that the congestion (the maximum over the load of all edges and inner nodes, induced by the read and write frequencies, divided by the bandwidth of the edge or inner node, respectively) is minimized. It is known [10] that this problem can be solved optimally in linear time, if inner nodes are allowed to hold copies of shared data objects. In our model, inner nodes correspond to buses and therefore cannot store copies of shared data objects. We show that this restriction increases the complexity of the placement problem drastically: It becomes NP-hard. On the other hand, the main contribution of our paper is an approximation algorithm with runtime O(|X|·|V|·height(T) · log (degree(T)) that increases the congestion by a factor of at most 7.