Jeff Kelly and Dave Vellante broadcasted live from New York, from the MongoDB Days Event, talking with the big names in the industry. One of these key people is Jon Hoffman, Engineer Infrastructure Lead at Foursquare.

Jon leads the infrastructure team in New York, and one of the things they do is building higher-level services on top of Mongo itself. That involves building tools that the developers can use to interact with Mongo. Right now, another big project they tackle is splitting up their monolythic application server into multiple, smaller applications. “We’re developping this server-oriented architecture and we’re building a lot of tooling to make that possible,” said Hoffman.

• We chose Mongo because we knew we were going to have a lot of data

The biggest attraction for Hoffman at these events is meeting up with other big users of MongoDB. He is an early Mongo user, and Vellanted asked about the motivation behind the decision to work with this database. “We started using it about three-and-a-half years ago. At the time we were running a standard SQL engine, a popular open source version of SQL, and the reason we moved over to Mongo was knowing we were going to have a lot of data,” answered Jon Hoffman.

The exponential growth in data made it clear that it was going to become impossible to keep it all on one server, even with the biggest server available at the time. There was the need to shard the data. That can be done in many ways, either on a SQL engine, managing the data splits yourself, or with the help of a tool like Mongo, where a lot of that infrastructure and heavy-lifting is handled for you.

• Homegrown load balancing vs. inherent capabilities of Mongo

The manual process of splitting data across multiple servers, for example, involves picking unique identifiers and hashing data, as well as managing a lot of the complications yourself. There’s a lot of manual tooling involved, and it’s a real challenge. Mongo makes it easier to automate, as it has an auto built-in feature called Auto Balancing.