Deadlock-Free Routing in InfiniBand through Destination Renaming
ICPP '02 Proceedings of the 2001 International Conference on Parallel Processing
L-Turn Routing: An Adaptive Routing in Irregular Networks
ICPP '02 Proceedings of the 2001 International Conference on Parallel Processing
Effective Strategy to Compute Forwarding Tables for InfiniBand Networks
ICPP '02 Proceedings of the 2001 International Conference on Parallel Processing
A New Methodology to Computer Deadlock-Free Routing Tables for Irregular Networks
CANPC '00 Proceedings of the 4th International Workshop on Network-Based Parallel Computing: Communication, Architecture, and Applications
Effective Methodology for Deadlock-Free Minimal Routing in InfiniBand Networks
ICPP '02 Proceedings of the 2002 International Conference on Parallel Processing
Fast Routing Computation on InfiniBand Networks
IEEE Transactions on Parallel and Distributed Systems
Oblivious routing for fat-tree based system area networks with uncertain traffic demands
Proceedings of the 2007 ACM SIGMETRICS international conference on Measurement and modeling of computer systems
INFOCOM'96 Proceedings of the Fifteenth annual joint conference of the IEEE computer and communications societies conference on The conference on computer communications - Volume 2
Hi-index | 0.00 |
To realize a path in an InfiniBand network, an address, known as Local IDentifier (LID)in the InfiniBand specification, must be assigned to the destination and used in the forwarding tables of intermediate switches to direct the traffic following the path. Hence, path computation in InfiniBand networks has two tasks: (1)computing the paths, and (2 )assigning LIDs to destinations (and using the LIDs in the forwarding tables to realize the paths). We will refer to the task of computing paths as routing and the task of assigning LIDs as LID assignment Existing path computation methods for InfiniBand networks integrate these two tasks in one phase. In this paper, we propose to separate routing and LID assignment into two phases so as to achieve the best performance for both routing and LID assignment. Since the routing component has been extensively studied and is fairly well understood, this paper focuses on LID assignment whose major issue is to minimize the number of LIDs required to support a routing. We prove that the problem of realizing a routing with a minimum number of LIDs is NP-complete, develop a number of heuristics for this problem, and evaluate the performance of the heuristics through simulation. Our results demonstrate that by separating routing from LID assignment and using the schemes that are known to achieve good performance for routing and LID assignment separately, more effective path computation methods than existing ones can be developed.