Be General and Don’t Give Up Consistency in Geo-Replicated Transactional Systems

Turcu, Alexandru; Peluso, Sebastiano; Palmieri, Roberto; Ravindran, Binoy

doi:10.1007/978-3-319-14472-6_3

Cited by 18 publications

(13 citation statements)

References 17 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In Alvin [24], Turcu et al outline a commitment protocol comprising a concurrency control layer on top of partial order broadcast, using rotating leaders over a sequence of delivery slots, not dissimilar to Mencius [7]. Each slot has an exclusive leader; its failure requires an election before a value may be proposed in that slot, revealing Alvin's adversarial nature.…”

Section: A Parallels Outside Of Consensusmentioning

confidence: 99%

“…Atomic commitment protocols have perceptibly evolved in the same vein as their consensus and atomic broadcast counterparts, relying on process exclusivity and multiple distinct phases to attain a commit/abort agreement over a set of resource managers, and in some cases [24], on the deterministic ordering of transactions in a log.…”

Section: A Parallels Outside Of Consensusmentioning

confidence: 99%

See 1 more Smart Citation

Spire: A Cooperative, Phase-Symmetric Solution to Distributed Consensus

Koutanov¹

2021

IEEE Access

View full text Add to dashboard Cite

All existing solutions to distributed consensus are organised around a Paxos-like structure wherein processes contend for exclusive leadership in one phase, and then either use their dominant position to propose a value in the next phase or elect an alternate leader. This approach may be characterised as adversarial and phase-asymmetric, requiring distinct message schemas and process behaviours for each phase. In over three decades of research, no algorithm has diverged from this basic model, alluding to it perhaps being the only viable solution to consensus. This paper presents a new consensus algorithm named Spire, characterised by a phase-symmetric, cooperative structure. Processes do not contend for leadership; instead, they collude to iteratively establish a dominant value and may do so concurrently without conflicting. Each successive iteration is structured identically to the previous, employing the same messages and invoking the same behaviour. By these characteristics, Spire buckles the trend in protocol design, proving that at least two disjoint cardinal solutions to consensus exist. The resulting phase symmetry halves the number of distinct messages and behaviours, offering a clear intuition and an approachable foundation for learning consensus and building practical systems.

show abstract

Section: A Parallels Outside Of Consensusmentioning

confidence: 99%

Section: A Parallels Outside Of Consensusmentioning

confidence: 99%

Spire: A Cooperative, Phase-Symmetric Solution to Distributed Consensus

Koutanov¹

2021

IEEE Access

View full text Add to dashboard Cite

show abstract

“…In addition, its graph-based dependency linearization mechanism that is adopted to define the final order of execution of commands may easily suffer from complex dependency patterns. Instead, Alvin [11] avoids the expensive computation on the dependency graphs enforced by EPaxos via a slot-centric decision, but it still suffers from the same vulnerability to conflicts of EPaxos: a command's leader is not able to decide on a fast path if it observes discordant opinions from a quorum of nodes. That is not the case of CAESAR, whose fast decision scheme is optimized to increase the probability of deciding in two communication delays regardless of discordant feedbacks.…”

Section: Related Workmentioning

confidence: 99%

“…To overcome this limitation, protocols aimed at allowing multiple nodes to operate as command leaders at the same time [9], [10], [11] have been proposed. Such solutions provide implementations of Generalized Consensus [12], a variant of Consensus that agrees on a common order of non-commutative (or conflicting) commands.…”

Section: Introductionmentioning

confidence: 99%

Speeding up Consensus by Chasing Fast Decisions

Arun

Peluso

Palmieri

et al. 2017

2017 47th Annual IEEE/IFIP International Conference on Dependable Systems and Networks (DSN)

Self Cite

View full text Add to dashboard Cite

This paper proposes CAESAR, a novel multi-leader Generalized Consensus protocol for geographically replicated sites. The main goal of CAESAR is to overcome one of the major limitations of existing approaches, which is the significant performance degradation when application workload produces conflicting requests. CAESAR does that by changing the way a fast decision is taken: its ordering protocol does not reject a fast decision for a client request if a quorum of nodes reply with different dependency sets for that request. The effectiveness of CAESAR is demonstrated through an evaluation study performed on Amazon's EC2 infrastructure using 5 geo-replicated sites. CAESAR outperforms other multi-leader (e.g., EPaxos) competitors by as much as 1.7x in the presence of 30% conflicting requests, and single-leader (e.g., Multi-Paxos) by up to 3.5x.

show abstract

“…To solve the fairness problem, a node must be allowed to choose which other nodes to communicate with based on the performance of its network links, as in the EPaxos [6] and Alvin [7] algorithms, both of which do not employ a unique leader, and allow any node to choose the closest (in terms of latency) set of nodes to reach agreement. EPaxos and Alvin are based on a novel but intricate algorithmic idea first introduced in EPaxos, which is difficult to generalize to devise other algorithms with different performance characteristics.…”

Section: Introductionmentioning

confidence: 99%