Why a SQL Pioneer Backs NoSQL: Insights from Donald Chamberlin

Donald Chamberlin, one of the co‑authors of the relational database language SQL, recently voiced strong support for the NoSQL movement, arguing that new web‑scale applications require scalability and performance that early relational systems were not designed to deliver.

He explains that NoSQL databases come in many forms, each addressing different challenges. Document databases such as MongoDB, Couchbase, and Amazon DynamoDB use open JSON standards to bypass relational schema constraints. Key‑value stores like Aerospike, Memcached, and Redis provide flexibility and speed, often serving as caches for internet applications. Graph databases such as Neo4j and TigerGraph model relationships as nodes and edges.

Chamberlin notes that many modern relational systems have begun to adopt NoSQL‑like features, for example supporting JSON documents in Oracle and PostgreSQL, and offering graph‑style queries.

He emphasizes that to achieve higher performance, data is frequently distributed across clusters, accepting eventual consistency rather than strict immediate consistency, which is acceptable in highly scalable environments.

Now semi‑retired from IBM, Chamberlin serves as a technical advisor for Couchbase, advocating a new query language designed to overcome the "impedance mismatch" between application languages and database structures.

University of California, San Diego professor Yannis Papakonstantinou introduced SQL++, a language intended to bridge the gap between JavaScript’s JSON‑centric data model and traditional SQL. Couchbase and AWS have adopted this language, with AWS branding it as PartiQL.

In a 2019 paper, Chamberlin compared SQL++ with SQL:2016, presenting example queries to illustrate their differences. He stresses that SQL++ is not a replacement for SQL but a complement, and that SQL will continue to exist because a large portion of business data remains in relational databases with robust, open‑source implementations such as MySQL, PostgreSQL, and SQLite.

"SQL will not disappear," he says. "Relational databases and the SQL language will accompany us for a very long time."