SDP: Scalable Real-time Dynamic Graph Partitioner

TL;DR

This paper introduces SDP, a scalable real-time dynamic graph partitioning method that reduces communication costs and balances load efficiently for evolving large graphs in streaming environments.

Contribution

The paper presents a novel streaming-based dynamic graph partitioning algorithm with new vertex assignment, communication-aware balancing, and scaling techniques.

Findings

Achieves up to 90% reduction in communication cost.

Balances load dynamically with 60-70% efficiency.

Reduces execution time significantly during partitioning.

Abstract

Time-evolving large graph has received attention due to their participation in real-world applications such as social networks and PageRank calculation. It is necessary to partition a large-scale dynamic graph in a streaming manner to overcome the memory bottleneck while partitioning the computational load. Reducing network communication and balancing the load between the partitions are the criteria to achieve effective run-time performance in graph partitioning. Moreover, an optimal resource allocation is needed to utilise the resources while catering the graph streams into the partitions. A number of existing partitioning algorithms (ADP, LogGP and LEOPARD) have been proposed to address the above problem. However, these partitioning methods are incapable of scaling the resources and handling the stream of data in real-time. In this study, we propose a dynamic graph partitioning method called Scalable Dynamic Graph Partitioner (SDP) using the streaming partitioning technique. The SDP contributes a novel vertex assigning method, communication-aware balancing method, and a scaling technique to produce an efficient dynamic graph partitioner. Experiment results show that the proposed method achieves up to 90% reduction of communication cost and 60%-70% balancing the load dynamically, compared with previous algorithms. Moreover, the proposed algorithm significantly reduces the execution time during partitioning.