NoSQL simply isn't hip anymore

The once red-hot database technology is losing its luster, as NoSQL reaches mass adoption

NoSQL simply isn't hip anymore
Credit: Shutterstock

This fall at Couchbase Live NYC there was a weird buzz. Ironically, it wasn’t that Couchbase 4.0 went to general availability with a new storage engine and, finally, the ability to run dynamic queries in SQL. The weird buzz was the lack of buzz. NoSQL isn’t cool any more. The hipster hackers have moved on.

This isn’t only Couchbase, but the whole NoSQL sector. When I was at Couchbase Live, my co-workers were in Chapel Hill, N.C., running Hadoop, MongoDB, and Spark training sessions at the Data Modeling Zone. Spark was, of course, the best attended -- yet even the Hadoop session was better attended than the MongoDB session (although the few people we talked to at the MongoDB session were serious prospective customers as opposed to hipster hackers).

The moment the buzz wore off NoSQL databases was right about the time they reached mass adoption. According to Thomas Vidnovic, a solution architect at Marriott International, Marriott hotels decided to finally ditch their mainframe and go to a Java/JSON/Couchbase-based solution. This was at least in part because they wanted to go to the cloud.

Marriott isn't the only franchise to decide that NoSQL is the on-ramp to the cloud. CenterEdge is a company that handles point-of-sale and management of franchises (like the one you took your kid to in order to jump on trampolines). The company was founded in 2004, when almost all of these sorts of companies used on-prem software. According to developer Brant Burnett, when CenterEdge started, the company used PC Anywhere to diagnose problems for its customers. Now it uses a combination of on-prem and cloud -- and sees a near-term future when every client is cloud-only.

Part of this change is the growing reliability and affordability of Internet connections, and part of this is that customers increasingly expect to be able to book everything online. Meanwhile, customers are opening multiple sites and want to manage, monitor, and analyze their performance centrally. They want to run cross-site promotions and reward key employees. A combination of cloud-based centralized management, BI-style dashboards, and analytics-plus-Couchbase are the key to their solution.

Cloud certainly isn’t the only driver behind NoSQL; scale continues to be important. I also spoke with Tony Selke, director of product engineering at HomeNet Automotive. HomeNet is part of the Cox Automotive conglomerate, which is mainly centered on services for car dealers. HomeNet originally deployed Couchbase four years ago because the company needed to scale beyond the 6 million cars it was designed for. Originally HomeNet brought it in because it had a Windows-based installer, and the company could offload operations to Couchbase as a cache. HomeNet has since made Couchbase the system of record as data has moved faster and it needed parallelism.

According to Selke, “One of the reasons that we have been investing more heavily in the Couchbase is that this past year we bought the biggest piece of hardware that Dell offers and we put two or three Fusion IO cards in it for storage because we needed to have that kind of throughput. But the reality is that barring some sort of large hardware advance, there is a cap and an expense to doing the vertical scaling.”

To be clear, CenterEdge plans to continue to run its own data centers -- and to identify specific parts of the pipeline that can be offloaded into the cloud and to use the cloud for peak operation.

NoSQL databases like Couchbase are simply part of our everyday reality. Ultimately, whether your volume forces you to rethink your architecture, you want to move to the cloud easily, or you're facing serious competitive pressure, you’re going to run something other than Oracle or SQL Server for at least part of your operation. The choice is so obvious -- NoSQL isn't hip anymore.

From CIO: 8 Free Online Courses to Grow Your Tech Skills
View Comments
Join the discussion
Be the first to comment on this article. Our Commenting Policies