BOSTON--(BUSINESS WIRE)--Cambridge Semantics, the leading provider of big data management and enterprise analytics software, today released its 2019 AnzoGraph Benchmark Report that arms organizations to perform deeply connected data analysis with improved computation. The report finds that AnzoGraph – a graph OLAP (Online Analytical Processing) database offered by Cambridge Semantics – can outperform for analytics and complement OLTP (Online Transactional Processing) databases. AnzoGraph allows administrators to scale for performance and data size to address analytics, data science and machine learning needs.
The report captures the difference between graph OLTP and graph OLAP databases by comparing execution speed of analytical style queries on a popular graph OLTP database vs. AnzoGraph by Cambridge Semantics, a graph OLAP database. The findings show an expectation of scale with AnzoGraph, which can load and perform analytics on a 100 Billion Triples benchmark, tackling 22 GHIB queries in 3-½ minutes, with the longest running query at about sixty seconds.
“Graph algorithms and analytical queries are important aspects of the unique value of graph databases and they provide a unique focus on data relationships, or the connections in data,” said Steve Sarsfield, Vice President, Product for AnzoGraph. “The 2019 AnzoGraph Benchmark report reveals that graph OLAP-style databases like AnzoGraph can help organizations better manage heavy workloads and scale to handle trillions of triples or more with elegance.”
Key findings from the benchmark report include:
- The graph database market is rich with OLTP solutions that read and write single records fast. When scaling, OLTP databases allow for increased read and write transaction speed, but scale poorly for data loading and complex analytics.
- Graph OLAP databases read millions of records fast, but do not read/write individual records as fast as OLTP. Graph OLAP databases like AnzoGraph scale differently and are better suited to provide deep query computation, multi-hop performance and analytics across the full corpus of data.
- Often, organizations need both types of databases to create a solution that includes transactional performance as well as analytics that can scale to handle big data, data science and machine learning workloads.
- As expected, AnzoGraph was significantly faster on the analytical queries than the tested OLTP database (Neo4J) on a single node; AnzoGraph took about sixty seconds total to complete the entire set of the queries, while Neo4J took 3.8 hours.
“For this workload of 22 queries, we could complete the analytical benchmark queries 217 times with AnzoGraph for every 1 completion of Neo4J. We think this performance gap is typical of most OLTP graph databases. When it comes to handling massive amounts of data and deep analytical queries, an OLAP-style database like AnzoGraph is unique in its ability to speed through queries at big data scale,” added Sarsfield.
To Learn More:
The 2019 AnzoGraph Benchmark report will be highlighted in an upcoming webinar entitled, “AnzoGraph Graph Database Benchmarking - the Numbers Tell the Story,” on January 30, 2019, 2 pm EST / 11 am PST. The GHIB benchmark, based on the TPC-H data set and queries, is only one of the recent benchmarks comparing graph OLTP to OLAP. The webinar will also cover benchmarks like the LUBM Test and Graph 500 data set for graph algorithm performance. Even some independent reviewers like Wallscope have devised tests and those results will be shared in this webinar.
About Cambridge Semantics
Cambridge Semantics Inc., The Smart Data Company®, is a big data management and enterprise analytics software company that offers a universal semantic layer to connect and bring meaning to all enterprise data. The company offers two award winning products: Anzo for Enterprise Knowledge Graphs and integrated analytics and AnzoGraph, a graph analytics database.
Cambridge Semantics is based in Boston, Massachusetts.
For more information visit www.cambridgesemantics.com or follow us on Facebook, LinkedIn and Twitter: @CamSemantics.
