To bring together all the social networks of its members, Glowbl transforms their data into graphs using Neo4j

The Challenge

Bringing together all possible social networks, representing all the contacts in the form of graphs and managing these contacts and their interactions in real time.

The Strategy

Since its start-up in 2011, the actual model used by Glowbl was based on a representation of its users through graphs. Using an SQL database to represent and read graphs was almost impossible, even more so when it came to displaying complex requests. Glowbl soon felt the need for a graph database, which would allow them to update the user database in real time, and offer unlimited request capacity with very short response times.

Mathieu Labey, CEO and founder of Glowbl says: "The world is made of graphs. And IT is made up of lists. Neo4j, with its graph design, allows us to bring reality into the world of IT, otherwise a solution like ours wouldn't be possible."

The Solution

Once the decision was made to obtain a database of this kind, Glowbl had to find the right one. Among all the solutions on the market, they felt that Neo4j was the most complete and fully developed graph database.

Using Neo4j as the basis of the Glowbl solution offered unlimited possibilities for adding data in real time, producing messages for users and updating data in real time. Mathieu Labey, CEO and founder of Glowbl notes: "Neo4j allowed us to design a competitive platform which shook up standards and the rules of the game. And all in one third of the time. Currently, we have a more stable and powerful platform that we can rely on as the basis for our development and growth."

Two independent phases of development were needed to integrate Neo4j into Glowbl: the real-time graph of spatial relations between bubbles and the social graph of social and behavioural connections between users. For the real-time graph, the first phase was taken up with trials, prototypes and design. The implementation itself followed on from this. The server installation and the production deployment were relatively quick and simple due to the ease of use of Neo4j, which only requires a few specific configurations. Finally, Glowbl carried out a long-term optimisation phase, with a view to the use and performance it expected from this tool. Regarding the social graph, first it had to be designed, then initial testing was carried out using the Glowbl data. This was followed by the implementation and optimisation phase, with the support of a Neo4j consultant. Now Glowbl has completed the final phase of integration and adding requests to the graph in order to produce a set of user recommendations. Mathieu Labey says: "In both cases, the installation and configuration of the Neo4j graph servers turned out to be extremely standard and simple."

The Result

Mathieu Labey concludes: "For a concept such as ours, it is vital to offer real-time responses and to be capable of handling huge quantities of data. This is key to the real success or industrial failure of our model. In the 3 years we've been using Neo4j, not only has the data we process grown exponentially, but the Neo Technology solution has never had the slightest down time. We are therefore in a position to imagine a bright future and to continue to develop our functionality and the number of members." Today, Neo4j allows Glowbl to process an enormous amount of data with total peace of mind. For the real-time graph, the volume of Neo4j data depends on the real-time activity of the platform and is therefore proportional to the site usage. It also evolves according to the number of people connected, which makes it very dynamic. The flexibility of Neo4j allows us to identify the recipients of a message through the real-time reading of a graph when a user wants to talk with a group of people. For the social graph, the nodes on the graph represent Glowbl users and all their relations (social network connections, contact requests) as well as their use of Glowbl (using LiveStage, events, etc.). The graph is thus used to send recommendations to users based on their usage or their contacts. This means the volume of data is extremely fluctuating but the system can handle massive quantities of data.

Why choose Neo4j?

Glowbl chose Neo4j because it was the most complete and fully developed graph database, with the highest review ratings, of the 4 databases that the start-up considered at the time.

Download Case Study