Trillion edges benchmark: new world record beyond 100TB by TigerGraph featuring AMD based Amazon EC2 instances

Graph databases have become increasingly popular in recent years, as they are uniquely suited to handle complex, interconnected data. As data sets continue to grow, scaling up graph databases to enterprise scale becomes a daunting task. Enterprises with hundreds of millions of customers and hundreds of billions of transactions require scale-out across: data, compute, security, manageability, queries, and algorithms. The world’s largest companies hold their most valuable data in TigerGraph including JP Morgan Chase, Intuit, and Microsoft. The recent world record benchmark results for TigerGraph on AMD EPYC based cloud instances are absolutely thrilling for the world’s largest enterprises.

In this benchmark, TigerGraph was tested on its ability to handle big graph workloads in a real-world production environment in AWS using 72 R6a.48xlarge Amazon Elastic Compute Cloud (Amazon EC2) instances powered by 3rd Gen AMD EPYC CPUs. These instances were chosen for better price-performance, and the ability to scale well on memory-intensive graph queries. Exactly how big is big? It is 108 terabytes, 3x the previous world record. Customer environments frequently have daily or hourly incremental updates of tens of terabytes of connected data. The LDBC SNB BI workloads included two challenges: a micro-batch of insert and delete operations to modify the current graph, and handling complex network-shaped read queries that access a substantial portion of the data. These queries were designed based on choke points and challenging aspects of query processing, such as multi-joins that are both explosive and redundant, and expressive pathfinding; interested readers can check the academic paper on this benchmark here.

TigerGraph successfully processed complex, deep-link OLAP-style queries on a massive graph containing 217.9 billion vertices and 1.6 trillion edges. Remarkably, eleven of the data-intensive read queries returned results within 1 minute, while the remainder took between 1 and 10 minutes. These results underscore TigerGraph’s exceptional ability to manage substantial graph workloads at an unprecedented level of scale and AMD EPYC’s ability to handle memory-intensive graph queries.

In comparison to other graph database and relational database providers, TigerGraph is in a league of its own for its analytical and operational capabilities with large-scale graphs spanning 100+ machines. Combined with performance and scaling capabilities of AMD EPYC instances on AWS, this provides an ideal solution for customers looking to extract maximum value from their data. No other technology can achieve this record. Please refer to this FDR for detailed reporting.

TigerGraph’s capability to scale in graph analytics and business intelligence on graph-structured data up to hundreds of terabytes is a game-changer in the world of big data. As a result, TigerGraph is now the go-to solution for enterprises seeking to analyze intricate, interrelated data on a massive scale. Contact us to learn more https://tiger-staging.bachoodesign.com/contact