Are there any NoSQL standards emerging?

Question

As with most new technologies after a while a standard emerges.

Is there anything cooking for NoSQL?

Answer 1

The whole point of NoSQL is that there are no standard solutions. Every data storage problem is different, and you need to choose the data storage technology that is appropriate for your specific problem and not the one that is "the standard".

That's the whole premise of "Not Only SQL".

Take ACID (here's a pieve of advice you never thought you'd get on StackOverflow, or really anywhere after 1987 :-) ), for example. There is a wide array of problems which don't need ACID guarantees. For those problems, ACID is overkill. Overkill that translates into wasted I/O, wasted CPU cycles, wasted performance. Which means wasted heat and wasted energy, which in turn means wasted money on electrical and utility bills.

Some problems only need weaker forms of those guarantees. For example, for a wide array of web applications the so-called eventual consistency is plenty enough. Other problems need higher guarantees than what SQL-style ACID provides.

So, some NoSQL databases don't have ACID guarantees or only have them in a weaker form. Some can turn them on and off on a per-DB basis. Some can turn A, C, I and D on and off individually on a per-DB basis. Some can not only turn A, C, I and D on and off individually, they can finetune them on a sliding scale. Some can even do that on a per-query basis.

If you have hierarchical data, store it in a hierarchical database. If you have graph data, store it in a graph database. If you have key-value data, store it in a key-value database. If you have semi-structured document data, store it in a document database. If you have semantic RDF data, store it in a triple database. If you build a data warehouse, store it in a column database. And if you have relational data, then, by all means store it in a relational database. (But only if you actually have relational data!)

Answer 2

There is no single standard NoSQL solution, as Jörg explained (+1). The term NoSQL covers a wide array of database types, each tailored for a specific data domain.

Ayende's That No SQL Thing series takes a look at some of the mainstream NoSQL solutions and highlights the strengths and weaknesses of each type. He discusses the following:

• Key/value stores
• Column-family stores
• Document databases
• Graph databases

You can think of these different types as standards within NoSQL. Just remember that each of them is specialized for certain data storage problems. There's no "one size fits all" solution: all of them will continue to exist.

Answer 3

Some people have contemplated about standards for document db's: http://nosql.mypopescu.com/post/731261002/a-common-nosql-query-language .

However key-value-stores and document db's don't do joins and that means that their query languages are simple and easy to learn. There is less need for a common language like SQL.

However .NET developers can use LINQ to access document db's MongoDB and RavenDB, and some people are developing a LINQ provider to document db CouchDB: http://github.com/sinesignal/ottoman . LINQ isn't a NoSQL standard but a standard for everything that is related to data. You can use it to talk to a relational database or an xml file too.

Graph databases are very different from key-value-stores and document db's. I don't think you can unite them in one standard. I really don't know if it is possible to develop a LINQ provider for a graph database. I guess not but I'm not sure.

Answer 4

Some NoSQL product supports SQL or a super set of it. This is the case of OrientDB, a document-graph nosql dbms with the support of SQL. It's released under Apache 2 license.

Furthermore it can export document in JSON format (you can export/import the entire database in JSON). Other NoSQL products read/write JSON.

bye, Lvc@