Paper 2020/1466

Load Balancing for Sharded Blockchains

Naoya Okanami, Ryuya Nakamura, and Takashi Nishide

Abstract

Sharding is an approach to designing a highly scalable blockchain. A sharded blockchain achieves parallelism by dividing consensus nodes (validators) into groups called shards and making them process different transactions in each shard. In this paper, we economically analyze users’ behavior on sharded blockchains and identify a phenomenon that users’ accounts and smart contracts eventually get concentrated in a few shards, making shard loads unfair. This phenomenon leads to bad user experiences, such as delays in transaction inclusions and increased transaction fees. To solve the above problem, we propose a load balancing framework in sharded blockchains in which accounts and contracts are frequently reassigned into shards to reduce the difference of loads between shards. We formulate the contract reassignment as an optimization problem and present the algorithm to solve it. Further, we apply the framework to an existing sharding design (Ethereum 2.0) and modify the protocol to do load balancing. Finally, we simulate the protocol and observe smaller transaction delays and fees.

Metadata
Available format(s)
PDF
Category
Applications
Publication info
Published elsewhere. Minor revision. International Conference on Financial Cryptography and Data Security
DOI
10.1007/978-3-030-54455-3_36
Keywords
ShardingBlockchainLoad balancingGame theoryHeuristicsSimulated annealing.
Contact author(s)
naoya okanami @ layerx co jp
ryuya nakamura @ layerx co jp
History
2020-12-21: revised
2020-11-24: received
See all versions
Short URL
https://ia.cr/2020/1466
License
Creative Commons Attribution
CC BY

BibTeX

@misc{cryptoeprint:2020/1466,
      author = {Naoya Okanami and Ryuya Nakamura and Takashi Nishide},
      title = {Load Balancing for Sharded Blockchains},
      howpublished = {Cryptology {ePrint} Archive, Paper 2020/1466},
      year = {2020},
      doi = {10.1007/978-3-030-54455-3_36},
      url = {https://eprint.iacr.org/2020/1466}
}
Note: In order to protect the privacy of readers, eprint.iacr.org does not use cookies or embedded third party content.