Sharding is a promising technique for addressing the scalability issues of blockchain. It divides the $n$ participating nodes into $s$ disjoint groups called shards, where each shard processes transactions in parallel. We investigate scheduling algorithms for the blockchain sharding systems, where each transaction resides in a shard of the communication graph and attempts to access accounts at possibly remote shards. We examine batch scheduling problems on the shard graph $G_s$, where given a set of transactions, we aim to find efficient schedules to execute them as fast as possible. First, we present a centralized scheduler where one of the shards has global knowledge of transactions to be processed. For general graphs, where the transaction and its accessing objects are arbitrarily far from each other with a maximum distance $d$, the centralized scheduler provides $O(kd)$ approximation to the optimal schedule, where $k$ is the maximum number of shards each transaction accesses. Consequently, for a Clique graph where shards are at a unit distance from each other, we obtain $O(k)$ approximation to the optimal schedule. We also get $O(k \log s)$ approximation for Hypercube, Butterfly, and $g$-dimensional Grid, where $g=O(\log s)$. Next, we provide a centralized scheduler with a bucketing approach that offers improved bounds for special cases. Finally, we provide a distributed scheduler where shards do not require global transaction information. We achieve this by using a hierarchical clustering of the shards and using the centralized scheduler in each cluster. We show that the distributed scheduler has a competitive ratio of $O(\mathcal{A_\mathcal{CS}} \log ^2 s)$, where $\mathcal{A_\mathcal{CS}}$ is the approximation ratio of the centralized scheduler. To our knowledge, we are the first to give provably fast transaction scheduling algorithms for blockchain sharding systems.
翻译:暂无翻译