Advancements in Distributed Storage and Replication Protocols

The field of distributed storage and replication protocols is witnessing significant advancements, driven by the need for efficient, reliable, and scalable solutions. Recent developments are focused on overcoming the limitations of traditional approaches, such as fixed-sized elements, single-point failures, and high latency. Researchers are exploring hybrid protocols that combine the benefits of different techniques, such as replication and erasure coding, to achieve better performance, availability, and storage costs. Another area of innovation is the design of leaderless protocols that can maintain non-zero throughput even in the presence of failures, and simplify the failure-recovery process. Noteworthy papers include: HyRES, which proposes a hybrid replication and erasure coding approach to data storage, resulting in additional design flexibility and better potential performance. LARK, which introduces a synchronous replication protocol that achieves linearizability while minimizing latency and infrastructure cost, and provides significant availability gains. EPaxos*, which presents a simpler and correct variant of Egalitarian Paxos, with a simpler failure-recovery algorithm and generalizes the protocol to cover a wider range of failure thresholds.

Sources

Rateless Bloom Filters: Set Reconciliation for Divergent Replicas with Variable-Sized Elements

HyRES: A Hybrid Replication and Erasure Coding Approach to Data Storage

LARK - Linearizability Algorithms for Replicated Keys in Aerospike

Making Democracy Work: Fixing and Simplifying Egalitarian Paxos (Extended Version)

Built with on top of