On: The Next 50 Years of Databases (2015)

[FillardMillmore]

I certainly think that NoSQL databases are only going to grow more popular in the future for more complex and specialized use cases. I don't see traditional RDBMS systems going away any time soon, but I don't know if they'll maintain dominance in perpetuity.

> The relational model will still dominant for most applications...Likewise, SQL (or some dialect of it) will remain the de facto language for interacting with a DBMS

The author does believe RDBMS systems will continue to dominate for the next 50 years. I have no particular reason to cast serious doubt about that, but it will certainly be interesting to see what the role and prominence of NoSQL databases is in that future.

[takeda → FillardMillmore]

I believe that's what the NewSQL is trying to tackle.

[threeseed → takeda]

NoSQL means a lot of categories of database e.g. graph, wide, columnar, document.

And so not everything is relational which is what SQL is for.

[Izkata → threeseed]

"NewSQL", not NoSQL, is apparently taking relational databases and adding on the features that drive people to NoSQL.

(This is also the first time I've heard that term, but it seems to be several years old and that's roughly what I got from a few quick searches)

[takeda → Izkata]

Google Spanner, CockroachDB etc are in that category.

[jimbokun → FillardMillmore]

The NoSQL databases will be query-able with SQL and start supporting ACID transactions, and then they won't be NoSQL anymore.

[FillardMillmore → jimbokun]

All of them? Why would a NoSQL database like Neo4j (Graph database) forfeit its declarative ASCII-like query language for SQL? What would the benefit be?

[threeseed → FillardMillmore]

Details like that aren't important because the right answer to every single data problem is "SQL" /s

Even if the data problem is graph, document, time series, wide table, EAVT etc.