6 ms·
Ask HN: What are some novel and niche database paradigms?
Outside of nosql, relational, graph, and timeseries are there any database paradigms that are defined? I’d like to explore the niche cases of database needs that aren’t so straightforward or needed by the many.
- haney 7y agoI'm not sure that it's a completely different paradigm but Datomic (https://www.datomic.com/ https://www.datomic.com/) is a pretty unique database if you're interested in thought provoking design. Here's a cool video of Rich Hickey (the creator of Datomic and Clojure) talking about the underlying data structures and design decisions: https://www.datomic.com/ https://www.datomic.com/
- z3ugma 7y agoYou should check out M, a database with a built-in programming language that was one of the original nosql transactional databases in the 60s. YottaDB is a modern company proving an M db. https://learnxinyminutes.com/docs/m/ https://learnxinyminutes.com/docs/m/
- fulafel 7y agoDatabase as a term and concept has a lot of historical baggage and whether something is considered a "database" or not has a large marketing component. The traditional function of minimizing slow disk accesseses in the absence of reasonable sized caches is largely gone as an interesting research area today. You may want to look at ways of storing data in memory and/or distributed systems generally. For example, asset data storage and retrieval in games, succinct data structures, systems like etcd. Another tangential advice: look at some of the recent published research. Eg ACM SIGMOD proceedings. The landscape tends to have relatively shortlived fads that leave behind them a narrow tail of applications that are actual fits for the idea in question. (Industry examples: "big data analytics" -> Hadoop, "web scale online db" -> MongoDB, etc).
- segmondy 7y agoNoSQL is big, anything that's not relational is NoSQL. So key/val, document, column. There's also object oriented database. This probably covers almost what you're looking for - https://en.wikipedia.org/wiki/NoSQL https://en.wikipedia.org/wiki/NoSQL
- giberti 7y agoImmutable ledgers (blockchain) and GIS (mixed systems) and search indexes (elastic) all come to mind as more niche databases. Some of these uses can be modeled in a variety of engines.
- elamje 7y agoIf you want a good high level overview I recommend Designing Data Intensive Applications by Kleppmann. You will walk away with a good understanding of the tradoffs of each paradigm.
- probinso 7y agoDeductive databases are pretty neat
- probinso 7y agoalso probabilistic query systems like BayesDB https://github.com/probcomp/bayeslite https://github.com/probcomp/bayeslite
- amirouche 7y agoVersioned databases especially for tracking provenance and lineage for data science models might become of thing. See Apple's Entity Store [0] or my take on it [1] [0] https://www.youtube.com/watch?v=16uU_Aaxp9Y https://www.youtube.com/watch?v=16uU_Aaxp9Y [1] https://github.com/awesome-data-distribution/datae https://github.com/awesome-data-distribution/datae