DataStax shows off DSE Graph at the DataStax Summit
DataStax shows off DSE Graph at the DataStax Summit

At the DataStax Summit in London this week there has been a lot of focus on the launch last week of DataStax Enterprise (DSE) Graph. The focus started with CEO Billy Bosworth telling the audience that it realised the volume of data and the complexity of relationships within that data was causing users problems. While DataStax had looked at graph technology itself it realised that the better solution last year was to acquire Aurelius who already had a working product but needed to way to scale for the market.

Is relational technology dead?

Matthias Broecheler, Director of Engineering for DSE Graph, DataStax
Matthias Broecheler, Director of Engineering for DSE Graph, DataStax

We spoke with Matthias Broecheler, Director of Engineering for DSE Graph at DataStax, about graph technology, what problems it solved and will customers really abandon their relational database technology to take on something new. The latter is particularly important as IT departments have heavy investments in relational technology and skills that they won’t want to abandon. On top of that we’ve been hearing that relational is dead ever since object technology appeared in the 1980’s. Since then there have been several other technologies appear, yet relational still rules our data environment.

Broecheler smiled as I was asking the question and said: “We are not advocating moving away from relational technology. What we are doing is getting customers to look at where they are feeling the pain. Many of those customers have tried relational first and discovered the pain is too large. This is often when they realise the complexity of what they are trying to do is no longer manageable.”

What does Graph technology deliver?

It’s the question I hear a lot from people and it was interesting to listen to people at the DataStax Summit who, even after the keynote, were still struggling with it. According to Broecheler it’s simple: “It’s about providing a view of other possibilities from the data.” We asked for a more specific example so Broecheler gave us one.

“Look at the problem from a telco’s viewpoint. They want to understand everything they can about a customer when they call the support desk.” At this point we suggested that this was something that had been going on since the early 2000’s. At that time we saw CRM systems inside call centres evolve to give the call handler not just a view of the person calling in but also access to finance records and information on everything they were paying for. Some systems even introduced a ‘churn’ button to put the call handler in customer save mode.

“We go further than that” Broecheler said. “Using the data the network holds on a person we can look at who they call regularly and whether they are on the same network. If the telco loses that person there is a chance they could also lose those people in their personal network that they also talk to when they tell them about a bad experience.”

This is an interesting extension of the use of data and Broecheler pointed out that writing a relational query to find and return that data would be incredibly complex and very resource hungry. This is where he believes that Graph databases and the use of Graph technology has its sweet spot. Part of that attraction is undoubtedly the visualisation and the example given by Broecheler is easy to see when he demonstrates it.

We were interested as to how far customers had gone and how easy it was for them to get it. For example, were companies also combining this approach with social media fire hoses to see the impact of a bad customer review? Perhaps they were using it to see what a customer said after talking to customer support teams?

“Not at the moment” Broecheler commented. “They are just working with the data that they have to get new insights.” We returned to the question about how they got it. “It is easy with telco’s as they already have complex customer systems,” responded Broecheler , “and all we are doing is allowing them to do more, faster and cheaper with their existing data.”

So how do companies get started?

“Some companies approach this problem from a data scientist approach” said Broecheler . “They want to know what data do they have to solve the problem. They go and talk to the different departments in the company and get the data which they then put into something new from which they can build a system. They build a prototype of a system and after a few iterations they then hand it over to the software product team who take all of that and create a working system for the users.”

There are several problems with this approach. The data is not always consistent, which creates problems in building queries across multiple data sources. Even when the queries work they are often incredibly complex, which means changes take a lot of time and the solution is not flexible. On top of that building a solution and deploying it can take a lot of resource to tune in order to meet client requirements.

To get around this Broecheler said: “We have a set of tools with DSE Graph that helps make it easier to get the data into DSE Graph.” We were interested in whether this was an unpublished API or perhaps they were capturing and replaying the existing log files from other databases. Broecheler shook his head saying: “No, we use a messaging approach not an API to write to but something you take data from.”

One of the things that DSE Graph uses is Apache Kafka. According to Broecheler this sits between the existing user front-end and their backend databases and simply captures the data. That data is then redirected into DSE Graph. The result is no need to dump existing technology, rewrite apps or change the way a company works. DSE Graph can just be deployed as an upgrade to deliver new ways of visualising the data.

Where do companies get skilled workers from?

Broecheler accepted that this was a challenge. “There is a massive need for education” he said. “We are rolling out the DataStax Academy in order to train and to enable people to think in terms of graphs.” The challenge for DataStax is that while it has some big customers and there are other companies in this space, Graph technology is likely to be yet another skill that IT people are going to be expected to self-teach.


Graph technology has some very interesting applications far beyond stopping a customer from churning away from a mobile or media company. Understanding the use of assets and to see new risks and patterns in security are just two of those that we will look at in a separate feature. What has to happen now is a wider discussion outside of tech companies as to how they can design, implement and benefit from the technology.

At the end of the DataStax Summit it was clear that delegates were getting the message. The question is can DataStax and its competitors drive Graph technology to the top of the “I want” list for end-users. One solution is to offer Graphing as a Service (GaaS) but this appears to be just a little way off yet. Whoever gets there first might just be able to grab some serious market share.


Please enter your comment!
Please enter your name here