Alan Giles: CTO at Boxever
Alan, good to talk to you for today’s Apache Cassandra Use Case. Can you give us an overview of what Boxever is all about?
Boxever is a Big Data analytics platform for the airline industry. Focused primarily on customer data – we enable airlines to capture data shopping and purchase behavior from every channel and combine that with external data such as information from social, user generated content and data from competitive websites to deliver a single customer view for every single traveller in the airline that is used to power our analytics, 1-1 marketing and personalized merchandising – all in real-time.
Myself and my fellow co-founders, Dave and Dermot, worked together at a travel technology company here in Dublin where we built and delivered multi-million-dollar e-commerce solutions to some of the largest airlines in the world. What we saw was that relatively few had good analytics and insight – particularly in the online channel. We felt that the airlines really were a long way behind other online retail verticals. We started Boxever to fix this problem.
We launched in Q4 2012 when we announced our seed round of $1M from leading Irish investors. We are live with a number of European carriers.
We’re a SaaS provider designed specifically for the travel industry, which means there are no long, expensive hardware and software procurement cycles, no consultants and no customization – just a monthly fee. Initial integration with an airline takes less than 4 weeks, so they get a great out-of-the-box experience with us.
Is your platform deployed on premise or in the cloud?
We’re on Amazon EC2 and are spread across three different availability zones.
What brought you to Cassandra and DataStax Enterprise?
We originally started with MySQL as the backend database, but we quickly found that MySQL just won’t scale for the things we need to do. So we’ve now made the total transition to DataStax Enterprise.
We take in so much data that we need something that can scale with the intensity of the data velocity we experience. For example, we consume all data relating to a user’s browsing behavior on airline sites. This includes all the various flight combinations that are suggested to a particular user along with what combination they eventually settled on. We report back to the airlines things like which of their offers and flight combinations are the best performing. That data alone can be massive to consume and analyze.
We’re using Cassandra to pull in all this data and more, and then we’re providing real-time analytics back to our airline customers about the things that matter to their business.
What are some examples of the real-time analytics you crunch through?
We do everything from determining what a particular user’s intent is, comparing that user against all other customers of an airline’s website in regard to behavior, their likes and dislikes, and producing analysis for an airline’s recommendation engine for what types of offers each user should be receiving at a particular time.
Any other particular technical aspects that caused you to go down the NoSQL route?
Our data model absolutely must be flexible, so this was another driver for us to go with something other than a traditional RDBMS.
Did you try other NoSQL products before settling on DataStax Enterprise?
Yes, we did evaluate other NoSQL options and looked especially hard at MongoDB. However, during our evaluation Cassandra demonstrated more positives such as better write latency, no single point of failure, and linear scalability.
Why did you go with DataStax Enterprise over just the community version of Apache Cassandra?
The big decider for us was the integrated Hadoop and Solr functionality.
Alan, sounds like you guys are doing some pretty interesting things – thanks for the time.