Couchbase's chief techie on squaring the schema-less circle

Interview Couchbase is a JSON document database with global users including PayPal, eBay, and travel distribution system Amadeus, companies with a combined revenue of nearly $40 billion. Its total annual income was $154.8 million, so why is such a small database outfit trusted by big biz?

We caught up with CTO Ravi Mayuram to find out about the history of the system which became a leader in the NoSQL movement. You can view the video below.

Mayuram argued that since relational systems were first designed in the 1970s, the IT landscape has changed. “Networks are faster, memory is cheaper, and there is a cloud-based consumption model. These were not available when the relational systems were built, so they are showing their age. We wanted to build something that would stand the test of time, just like relational systems did,” he said.

One of the significant departures Couchbase took from the relational approach to databases was to avoid designing the data schema up front. This move has attracted criticism from some database experts.

“Having the schema is when you really pay the price. If you have a schema, data is in a solid brick form. Whenever you want to move that, you have to sort of re-cast that into a new schema when you go from system to system,” he said.

Instead, Couchbase employs something called a late-binding schema, which are not tied to the underlying database objects that it references.

• Microsoft adds silicon muscle into latest Azure SQL database configs
• MotherDuck scores $47.5m to prove scale-up databases are not quackers
• Redis swallows RESP.app biz that made its database easier on developers
• Rows, columns, and the search for a database that can do everything

“You always need a schema — without that there is no structure — but a schemaless database removes that friction between front end and back end. There is no separate application schema and database schema; there is only one schema which is what the application defines,” he said.

But from Couchbase 7.0 it began to look a bit more like a relational database by offering multi-statement SQL transactions which cover what relational systems can offer from the standpoint of OLTP guarantees but also preserves performance, scale and flexibility, the company argued.

In the video, Mayuram talks listeners through these design choices and elaborates on how they change the role of developers and DBAs within tech teams. ®