Graph databases are distinguished by relationships. Users can query, for example, the connections that relate a customer to an account that he or she owns, or how many degrees of separation exist between Kevin Bacon and Audrey Hepburn.
Ontologies are a key underpinning, providing a data model to describe things in a database. Metadata and data quality matter so that appropriate conclusions are reached — just which Audrey Hepburn are we talking about anyway?
The Data About Graph Databases
Donna Burbank, the Managing Director at Global Data Strategy, recently presented a webinar that discussed recent developments in graph databases, and the results of the of the survey conducted for the DATAVERSITY® Trends in Data Architecture Report on which the topic was based.
Some findings of the survey are that:
- Only 12.7% of respondents were currently using a graph database. Almost a quarter of them, however, were planning to use a graph database in the future.
- Demand for graph databases is growing because businesses want the ability to flexibly create enterprise knowledge graphs, understand social media connections and accomplish fraud detection.
- When looked at as a group to augment relational databases, the use of key-value, document, columnar, and graph data stores is a significant number at 70.5%.
Get Going with Graph Databases
How can companies step themselves into the world of graph databases? Neo4j thinks it has an answer. It has been offering a Startup Program for startups with 19 employees or fewer; more than 650 startups with fewer than 20 employees took advantage of having free access to Neo4j Enterprise clusters. The company recently expanded its program to support startups with up to 50 employees and $3 million in annual revenue.
Companies whose applications are accepted now can have free access to Neo4j Enterprise Edition, which provides developers with the design, development, maintenance, and operational power for enterprise-class graph applications, the company says. Also part of the package is Neo4j Bloom, a graph exploration application available in the Neo4j Graph Platform for investigating and exploring graph data visually from different business perspectives. SMBs also have access to innovations driven by Neo4j Labs, such as optimized graph algorithms and machine learning libraries.
A Startup Story
One company taking advantage of the Startup Program is Quander, an early-stage startup that provides tools for organizations to track customer engagement and build experiences in physical spaces — brand experience, events and conferences, and retail.
Quander’s platform has always been based on traditional RDMS, and it also had dabbled with other database technologies like memory, document stores, and data warehouses. But it always found itself coming back to some kind of RDMS with support for JSON storage, because nothing else truly met its requirements.
Then, said Gavin Williams, CEO and co-founder, “we stumbled upon graph technology and were immediately infatuated by it. It answered most, if not all, of our needs for data storage, querying, and analytics.”
The company became involved in the Neo4j Startup program last December.
In the Q&A below, Williams discusses more about the value of graph technology for Quander, today and in the future.
DATAVERSITY (DV): What were your primary needs for getting involved a startup program like this one?
Gavin Williams: We use graph databases because they allow us to store information as we see it in the real world. This not only gives us the ability to answer the questions we have about our data today, but the questions we may have tomorrow without requiring us to write a ton of joins or refactor our database. Additionally, the rapid prototyping through Neo4j allows us to build new APIs quickly.
At the moment, graph is meeting the majority of our requirements. We’re at the point where we can pour millions of nodes and edges without any real disadvantages from what we currently have. We expect using Neo4j will accelerate our development to some degree as well as reduce the amount of time it takes us to add new analytical features to our platform because it’s so flexible.
We live in a day and age where it’s really easy to come up with an idea and deliver it. I’ve seen time and time again that the database behind a platform takes a backseat as part of the build process. Neo4j allows you to focus on building your application without worrying about whether your data store is going to fail or not meet your needs in the future. If you can model something in the real world, you can model it in Neo4j, and Neo4j can adapt to your product.
DV: How about cost savings?
Williams: There’s a really clear upgrade path for Neo4j for anyone getting to grips with it and building their applications. Most developers will start with the free edition to build their applications locally, which is easy to install on Windows, MacOS and Linux with either the desktop installer or Docker. After that, it’s pretty simple to spin up the community edition with Google Cloud and Kubernetes. Having access to the Enterprise Edition then allowed us to spin up a cluster for production which would have cost us a fair amount. For a startup, any money saved on software licenses can go towards the business’ growth.
DV: To date, how have you used graph technology to specifically enhance your applications?
Williams: We’re finding relationships between brands’ customer interests through interactions on our platform. This allows us to build a matcher service using our anonymized data so that Brand X and Brand Y can create partnered campaigns not just on the basis that they have similar customers — they can hyper-target their campaign based on common customer interests. Our platform also provides brands with the tools to then deliver their campaign in a retail or experiential environment and re-affirm the effect of that using Neo4j.
DV: Which of the Labs’ innovation projects have the most value to your business?
Williams: APOC (procedure library) and GrandStack (for building full stack apps), without a doubt. Without APOC we wouldn’t have been able to migrate over 10GB’s worth of sensor data from BigQuery to Neo4j. We’re only scratching the surface and are currently playing with virtual graphs. GrandStack has also allowed us to rapidly prototype and build our GraphQL API in a way that we’ve never seen before. We’ve ported it to TypeScript (as that’s our workflow).
Next on our list are graph algorithms; we’re working with one client in particular who is very interested in community detection to serve better product recommendations in retail. We also want to use this to optimize content within our engagement apps automatically.
DV: How important and how helpful has the online community that supports the lab projects been for your company?
Williams: Very important and very helpful. We recently did a Hangouts on Air session with Neo4j to give back to the community. We hope to do more of these once our platform is live so we can contribute our learnings back to the growing community.
DV: When do you think you’ll become an alumni of the Startup Program?
Williams: September 2019. That’s when our new platform goes live and we can share all of our learnings with the community!
Image used under license from Shutterstock.com