Retailers Embrace Graph Databases
Graph databases that use nodes and properties to represent and store data appear to be making significant inroads in the U.S. retail market.
For example, graph database specialist Neo Technology of San Mateo, California, recently announced that large retailers including eBay and Walmart have adopted its Neo4j graph database for critical business applications. The auction site is using the tool to improve same-day delivery services while Walmart has applied graph database technology to monitor online shopping behavior.
Neo Technology attributes the growing mainstream adoption of graph databases to their ability to help solve previously intractable technology problems. For example, the company said, retailers can leverage graph databases to collect and analyze peer-to-peer information in order to get a better sense of what products consumers want and at what price points.
Citing statistics from the database monitoring website DB-Engines, Neo Technology said the popularity of graph database management systems has soared by more than 250 percent in the last year.
According to DB-Engines’ most recent rankings, relational databases retain their stranglehold on the market. Neo Technology is the leading graph database supplier in the rankings at No. 22. Other graph database products include the Titan distributed graph database developed by Aerelius and the Sparksee database from Sparsity Technologies.
Retailers are also embracing emerging graph databases as the source of customer data shifts from suppliers and manufacturers in the form of marketing data to user-generated product reviews and social media.
Neo Technology customers said these shifts in consumer behavior and the rise of social media have outpaced the capacity of conventional database technologies like MySQL to keep pace with the surge in consumer data.
In eBay’s case, significant growth and new services like same-day delivery prompted the switch to graph databases as a way to chart and guarantee the fastest delivery time. Hence, the graph database was added to eBay’s existing SOA software and services infrastructure to improve performance and overcome scaling issues.
Besides speed and agility, graph databases like Neo4j are also being touted for reducing by factors up to 100 the amount of code needed to process database queries.
Seeking to bolster its same-day delivery service, eBay acquired the London-based delivery service Shutl in October 2013 to speed customers deliveries. Shutl uses a network of local couriers to deliver orders and provides customers with application-programming interfaces used to integrate their business with the Shutl service.
Volker Pacher, a senior developer at Shutl, said it adopted the Neo4j graph database when faced with the problem of developing a new API for its growing eBay business. Same-day delivery options range from as little as 90 minutes to a one-hour window of the customer’s choice.
One challenge was that API response times were growing as more connected data was added, Pacher said. “Our fastest delivery [15 minutes] was quicker than our slowest query,” he joked.
“The performance in a graph [database] is relatively constant because the queries are localized,” Pacher explained. “In real-world terms, let’s say we decide to start deliveries in Germany. That won’t effect our query time here in the U.K. [because] the query time is local and the time it takes to execute the query to find a courier for a delivery is constant.”
Graph theory is based on the work of Swiss-born Austrian mathematician and physicist Leonhard Euler, who applied the theory to solve the “Seven Bridges of Koenigsberg” problem in the 18th century. The problem involved traversing each bridge exactly once. Euler showed why such a route could not be found.
Little did Euler know that his graph theory would become a basis for understanding computer networks.