Balanced Allocation: Patience is not a Virtue
John Augustine, William K. Moses Jr., Amanda Redlich, Eli Upfal

TL;DR
This paper introduces a simple, decentralized load balancing algorithm called FirstDiff[d] that achieves the improved maximum load performance of more complex algorithms like Left[d], with similar probe complexity.
Contribution
The paper presents FirstDiff[d], a new decentralized algorithm that combines simplicity with improved load balancing, matching the performance of more complex methods.
Findings
FirstDiff[d] requires at most d probes on average per ball.
FirstDiff[d] achieves maximum load close to that of Left[d].
Experimental results show FirstDiff[d] performs as well or better than Left[d].
Abstract
Load balancing is a well-studied problem, with balls-in-bins being the primary framework. The greedy algorithm of Azar et al. places each ball by probing random bins and placing the ball in the least loaded of them. With high probability, the maximum load under is exponentially lower than the result when balls are placed uniformly randomly. V\"ocking showed that a slightly asymmetric variant, , provides a further significant improvement. However, this improvement comes at an additional computational cost of imposing structure on the bins. Here, we present a fully decentralized and easy-to-implement algorithm called that combines the simplicity of and the improved balance of . The key idea in is to probe until a different bin size…
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
