Building global and scalable systems with Atomic Multicast

TL;DR

This paper advocates for atomic multicast as a key abstraction for building scalable, globally distributed systems, demonstrating its effectiveness through the design and evaluation of two modern online services.

Contribution

It introduces the use of atomic multicast for designing scalable distributed systems and provides practical implementations with performance evaluations.

Findings

Atomic multicast effectively supports large-scale geographically distributed services.

The proposed services demonstrate strong consistency and scalability.

Experimental results show acceptable latency and throughput in global deployments.

Abstract

The rise of worldwide Internet-scale services demands large distributed systems. Indeed, when handling several millions of users, it is common to operate thousands of servers spread across the globe. Here, replication plays a central role, as it contributes to improve the user experience by hiding failures and by providing acceptable latency. In this paper, we claim that atomic multicast, with strong and well-defined properties, is the appropriate abstraction to efficiently design and implement globally scalable distributed systems. We substantiate our claim with the design of two modern online services atop atomic multicast, a strongly consistent key-value store and a distributed log. In addition to presenting the design of these services, we experimentally assess their performance in a geographically distributed deployment.