Damon Feldman on multi-model databases

O'Reilly Podcast: Working with databases that go beyond traditional models.

By Jon Bruner
November 8, 2016
Elevated parking lot. Elevated parking lot. (source: Pexels)

In this podcast episode, I talk multi-model databases with Damon Feldman, solutions director at MarkLogic. If this conversation gets you interested in multi-model databases, be sure to check out his live webcast on December 1, 2016.

Learn faster. Dig deeper. See farther.

Join the O'Reilly online learning platform. Get a free trial today and find answers on the fly, or master something new and useful.

Learn more

“A multi-model database at its basic [level] is really what it sounds like: it’s a database that stores multiple models in terms of how it represents data,” says Feldman. They go beyond classic entity-relationship models to store graphs, JSON documents, triples, and perhaps binary content.

Multi-model databases aren’t necessarily huge installations with many models; Feldman says the most common multi-model installations have three models, though two-model databases are also widespread. “I’m seeing more and more databases start to pick up at least two models,” he says. “Sometimes there’s a model that doesn’t do everything you need to do, and there’s a known gap, so you add one more model.”

Feldman describes the model types most commonly found in multi-model databases:

  • Entity-relationship model, like in a SQL database
  • Document model, often XML or JSON
  • Text model, opting to store text in a database rather than in a different repository that would require integration
  • Graph or triple model: RDF or semantic data. Essential when a lot of entities are linked together
  • Key-value store: a very simple “black-box” cache

So, what kind of skills does it take to implement a multi-model database? Feldman says it’s best to focus on the skills that are relevant to each of the model types in the database: OWL for dealing with triple models, JSON and JavaScript for document stores, and third normal form for traditional relational models, for instance. As multi-model databases become more popular, Feldman says the best databases will begin to have multi-model capabilities built into their cores.

For more on multi-model databases, watch Damon Feldman’s live webcast on December 1, 2016.

This post is a collaboration between O’Reilly and MarkLogic. See our statement of editorial independence.

Post topics: Data science