On weighted balls-into-bins games
Theoretical Computer Science
How to Balance the Load on Heterogeneous Clusters
International Journal of High Performance Computing Applications
On cost-aware monitoring for self-adaptive load sharing
IEEE Journal on Selected Areas in Communications
Randomized load balancing strategies with churn resilience in peer-to-peer networks
Journal of Network and Computer Applications
IWOCA'10 Proceedings of the 21st international conference on Combinatorial algorithms
Tight bounds for parallel randomized load balancing: extended abstract
Proceedings of the forty-third annual ACM symposium on Theory of computing
From balls and bins to points and vertices
ISAAC'05 Proceedings of the 16th international conference on Algorithms and Computation
Initiating load balancing operations
Euro-Par'05 Proceedings of the 11th international Euro-Par conference on Parallel Processing
On weighted balls-into-bins games
STACS'05 Proceedings of the 22nd annual conference on Theoretical Aspects of Computer Science
Approximated two choices in randomized load balancing
ISAAC'04 Proceedings of the 15th international conference on Algorithms and Computation
Anonymous distribution of encryption keys in cellular broadcast systems
MADNES'05 Proceedings of the First international conference on Secure Mobile Ad-hoc Networks and Sensors
Journal of Discrete Algorithms
Cost aware adaptive load sharing
IWSOS'07 Proceedings of the Second international conference on Self-Organizing Systems
Balls-into-bins with nearly optimal load distribution
Proceedings of the twenty-fifth annual ACM symposium on Parallelism in algorithms and architectures
| Hi-index | 0.00 |
A standard load balancing model considers placing n balls into n bins by choosing d possible locations for each ball independently and uniformly at random and sequentially placing each in the least loaded of its chosen bins. It is well known that allowing just a small amount of choice(d = 2) greatly improves performance over random placement (d = 1). In this paper, we show that similar performance gains occur by introducing memory. We focus on the situation where each time a ball is placed, the least loaded of that ball's choices after placement is remembered and used as one of the possible choices for the next ball. For example, we show that when each ball gets just one random choice, but can also choose the best of the last ball's choices, the maximum number of balls in a bin is l{{\log \log n} \mathord{\left/ {\vphantom {{\log \log n} {2\log \phi+ 0(1)}}} \right. \kern-\nulldelimiterspace} {2\log \phi+ 0(1)}} with high probability, where \phi= ({{1 + \sqrt 5 )} \mathord{\left/ {\vphantom {{1 + \sqrt 5 )} 2}} \right. \kern-\nulldelimiterspace} 2} is the golden ratio. The asymptotic performance is therefore better with one random choice and one choice from memory than with two fresh random choices for each ball; the performance with memory asymptoticallymatches the asymmetric policy using two choices introduced by Vöcking. More generally, we find that a small amount of memory, like a small amount of choice, can dramatically improve the load balancing performance. We also investigate continuous time variations corresponding to queueing systems, where we find similar results.