Topology-Aware Consensus Protocol for Distributed Systems
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing consensus protocols in distributed applications face scalability issues due to reliance on centralized coordinators, leading to increased latency and limited performance as the number of participants grows, and are not parallel or topology-aware, which restricts their scalability in wide-area deployments.
Innovation Solution
A leaderless, parallel, and topology-aware consensus protocol that groups nodes by physical proximity, allowing consensus to be achieved in multiple rounds, with each round aggregating and ordering write requests across progressively larger groups until global consensus is reached among all nodes, eliminating the need for a centralized leader and enhancing scalability.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a centralized coordinator is used to service client requests and replicate state changes, then the system achieves consensus, but the processing load and network traffic concentrate at the centralized coordinator, causing increased latency and limited scalability
Solution Approach 1:
The patent divides the centralized coordinator role into multiple coordinator nodes that operate in parallel. Each coordinator handles a subset of client requests and state changes, distributing the processing load across the network. This segmentation eliminates the single point of congestion while maintaining consensus through coordinated interaction between multiple coordinators.
Solution Approach 2:
The patent introduces a hierarchical dimension to the consensus protocol, organizing nodes into groups with local coordinators that first achieve consensus within their groups, then coordinate with other groups. This multi-level hierarchy transforms the flat single-coordinator model into a structured system that reduces network traffic at any single point while achieving global consensus.
2Reliability
If the number of participants in the consensus protocol increases to improve fault tolerance, then the system becomes more reliable, but the performance degrades due to increased coordination overhead
Solution Approach 1:
The patent segments the participant population into multiple groups, each with its own coordinator. This allows the system to scale to more participants while maintaining manageable coordination overhead within each segment. Fault tolerance is achieved through the distributed nature of multiple coordinators and groups, while performance is preserved by limiting the coordination scope within each segment.
Solution Approach 2:
The patent implements a two-phase consensus approach where coordinators first achieve partial consensus within their groups, then coordinate with other groups for global consensus. This partial action approach allows the system to handle more participants by breaking down the full consensus problem into smaller, more manageable sub-problems that can be solved independently before being combined.
3Device complexity
If message dissemination is performed sequentially without parallelism or topology awareness, then the protocol is simple to implement, but scalability is limited in wide-area deployments with restricted network link capacities
Solution Approach 1:
The patent segments message dissemination into multiple parallel rounds, where each round disseminates messages to a specific subset of nodes. This segmentation allows simultaneous message transmission across multiple network paths, utilizing available network capacity more effectively. The segmented approach maintains relative implementation simplicity while dramatically improving scalability for wide-area deployments.
Solution Approach 2:
The patent introduces topology awareness into the message dissemination process by organizing nodes into groups based on their network positions. Messages are disseminated in a structured sequence through different hierarchical levels (intra-group, inter-group, and cross-datacenter rounds), transforming the flat sequential dissemination into a multi-dimensional process that exploits network topology to improve scalability.
Data Source
AI summary
Methods are provided for achieving consensus among an order in which write requests are received by various ones of a plurality of nodes in a distributed system using a shared data structure. The plurality of nodes are organized into groups of nodes and successively larger groupings of groups, based on physical proximity. A consensus protocol is used to achieve consensus among groups of nodes, and then among the groupings of groups of nodes in a logical tree structure up to a root level virtual node. Recovery from failure of all nodes in a group is supported.