In practice, documented-oriented databases have many intrinsic flaws. Querying a document database requires "navigating" the document. Queries must be less declarative and more navigational, resulting in tighter coupling between applications and the database. Document stores have each invented their own query languages, a regression in standardization that is unnecessary.
Though the SQL query language has many proprietary extensions, it has a well-specified core that makes applications and user skills portable across database systems. Popular document stores remove an important abstraction between physical storage of data and logical structure of data. This will be a substantial limitation on optimizations over the long term.
The relational system abstracts the physical arrangement of data from its logical schema, enabling relational systems to make the definition of the data they store queryable. A program can introspect the logical structure and constraints of a relational system. This powerful ability has led to a rich ecosystem of management and monitoring tools for relational systems.
Finally, relational systems, unlike document store databases, allow normalization of data (i.e, data can be factored so redundant information is stored only once) and constraints that enforce consistency (e.g., every order must reference a valid customer). Document stores enable neither, putting the full complexity of maintaining correctness on the application developer.
Shaking out a solution
Database vendors are acknowledging the adoption, popularity and the utility of the JSON format and are integrating support for JSON in to their traditional relational systems -- a best-of-both-worlds approach.
Adding the ability to store and query JSON records in the context of the traditional relational system gives developers the benefit of a powerful choice. What portions of my data should be abstracted -- and where are constraints and strict schema guarantees important to my application? Where does flexibility instead of consistency checks matter most? Additionally, these integrations add JSON query capability as a natural extension to SQL, enabling powerful queries across arbitrary sets of JSON records combined with data stored in traditional table cells.
In reality, it will be substantially easier for relational vendors to extend their systems to include JSON than it will be for the new document store vendors to retrofit their non-relational systems with the features lost in the move away from relational structures. Enormous value remains in relational structure, which has proven itself over the last 20 years, along with the rich ecosystem that surrounds relational databases.
So what's so special about JSON? It's a microcosm emblematic of the deep impact of the Web and Web-oriented programming models, even penetrating to the recesses of the data management tier. It has giving rise to new vendors catering to the preferences of modern developers. And ultimately, JSON will also teach relational databases useful new tricks.
New Tech Forum provides a venue to explore and discuss emerging enterprise technology in unprecedented depth and breadth. The selection is subjective, based on our pick of the technologies we believe to be important and of greatest interest to InfoWorld readers. InfoWorld does not accept marketing collateral for publication and reserves the right to edit all contributed content. Send all inquiries to newtechforum@infoworld.com.
This article, "How JSON sparked NoSQL -- and will return to the RDBMS fold," was originally published at InfoWorld.com. For the latest business technology news, follow InfoWorld.com on Twitter.