Solving distributed graph problems with DataStax

8th July, 2019

We talk to Denise Gosnell, Head of Global Graph Practice at DataStax

Denise Gosnell - Senior Manager and Head of Global Graph Practice at DataStax

In 2016, we proudly welcomed DataStax, leading provider of modern, scalable, distributed databases built on Apache Cassandra™, to our Technology Alliance.

Their graph database, DataStax Enterprise (DSE) Graph, continues to give enterprises the power to store, identify and analyze the hidden connections between billions of data points. Through seamless integration with our graph visualization toolkit technology, users uncover insights and bring those vast datasets to life. Together, we solve real-world challenges, from fraud detection to customer 360, understanding social networks to securing the Internet of Things.

We caught up with Denise Gosnell, Senior Manager and Head of Global Graph Practice at DataStax, to talk about graph challenges, graph visualization, and what the future holds.

Hi Denise, could you tell us about your current role at DataStax?

Hi there – sure. I lead a team that connects our global customer base with the DataStax product & engineering departments. Our goal is to simplify and keep solving massively scalable, complex distributed graph problems using DataStax products. It’s fascinating to see the awesome solutions our customers are building for their hybrid cloud environments.

To be honest, it’s a dream role that I couldn’t be more honored to be in.

How did DataStax Enterprise (DSE) Graph come about?

Enterprise customers wanted a graph database built in Cassandra™ – to give immense scalability and availability – in addition to having tight integrations with external tools for solving graph problems.

Their graph problems are multi-faceted and complex. You need at least three tools to tackle them properly:

  • Analytics – DSE Graph leverages Spark and graph frames, so you can run graph algorithms over the global topology of your graph.
  • Search – tight integration with search capabilities so you can use lexical/fuzzy search as your starting point for traversals.
  • Multiple views of the data – inside one storage engine we can provide multiple views of the data, so you can look at your data like a graph or as a Cassandra table.

What kind of problems are your customers trying to solve?

There are so many creative ways to use our technology, but the three main areas right now are:

  • Customer 360: where organizations need to integrate data from many silos to present a single view to their customers. We’re helping brands modernize and rethink their internal data architecture by integrating their data sources into a single graph view.
  • Pathfinding: massively distributed graphs require hundreds of machines holding vast stores of nodes and edges. Organizations want to find paths through hundreds of millions of nodes fast, and then use that model to find and explore similar paths
  • Something we call ‘unbound tree traversal’. It lets you run the kind of expressive queries that are difficult to maintain in a relational system, such as identifying the massive list of dependencies at each stage of a manufacturing production line.
Achieve a single consolidated view of all your data using our graph visualization technology
Achieve a single consolidated view of all your data using our graph visualization technology

Why is graph visualization so important?

We interpret our environment using visual perception. Visualizing connections is what draws us into graph analytics – it’s a much more tangible and understandable way to interact with the data.

Whenever you’re able to see an image of the relationships in your data, it creates that human connection that’s much more attainable than if you’re looking at rows and columns.

Having a solid story like DataStax has with Cambridge Intelligence and being able to deliver that for production applications is why companies choose the partnership we’re providing.

What makes a DSE Graph user choose our graph visualization technology?

Customers want to create visualizations with a custom UI, but they also want that fully-capable toolset delivering production-quality interfaces with their graph data. That’s why KeyLines is the number one pairing between DSE Graph and production applications.

Our partnership has been extremely strong by guiding customers through creating production-quality, front-end visualizations.

Which of our toolkit technology features are most valuable to your customers?

Great question. There are two main features that our customers tell me are really nice. First there’s the support for performing neighborhood expansion right out of the box.

Then there’s the ability to completely customize the look and feel of the visualization. Other vendors don’t give you that full customization all the way down, to changing colors as deep as you want to go. Being able to fully flush out the whole user experience/flow and the view of it is one of the reasons our customers turn to Cambridge Intelligence.

Custom-designed interfaces powered by the Cambridge Intelligence graph visualization toolkit technology
Custom-designed interfaces powered by the Cambridge Intelligence graph visualization toolkit technology

And finally, what does the future hold for DSE Graph and our toolkit technology?

There are so many fantastic ideas. It’d be great to come up with smart suggestions around supernode data modelling through matching and merging data. I’d love us to partner on using the UI to separate nodes with too many edges and present a clearer visualization of the data.

Intuitive node combining tackles some of the supernode challenge by reducing clutter and revealing insight in complex graph datasets
Intuitive node combining tackles some of the supernode challenge by reducing clutter and revealing insight in complex graph datasets

From a developer’s perspective, it’d be great for nodes merged or grouped at the front end to persist at the back end, so there’s tighter integration between the UI and the database.

Our latest DSE Graph version is all about building distributed graph adjacency lists into Cassandra™’s data model, so users can get the fastest and most optimized experience.

These are different types of data that previously couldn’t live in the same system. It’ll be very interesting to see users create multiple views of the same graph structure and visualize the results with Cambridge Intelligence toolkits.

Visualize your DataStax Enterprise Graph

Once you’ve requested a free trial of our graph visualization technology, follow this tutorial to see how easy it is to integrate with DSE Graph.

Read more blog posts about Integrations.