An apt database (and/or) framework suggestion needed

Question

Well, I've taken help of Google, Stackoverflow and whatever else I could find, did as much as I could, but it seems that I am unable to find out an exact answer! I have multiple queries, and I would love to have answers from the database-people as well as from the programmers and framework users.

From the programming languages, I know C/C++, Java and Python. I have undertaken a CMS project that would require frequent C's & R's of the CRUD. The project would have 50k users atleast. The head-to-toe of the project has been all figured out, and now I need to code it and make it live online.

Well, I want to use Neo4j as my database as its data representation model (nodes and relationships) is closest to the real project model. Now, neo4j has bindings for various languages, and one of them is Python (whose python bindings are very oldish, the jpype hasn't been updated since ages). I am thinking of going for some Java based framework, but then I leave this idea as I personally haven't heard much of java frameworks. But one of my partner tells me to go for Zend (PHP) as it has some kind of functionality that lets us execute Java code. Won't this slow the code? I mean executing one language's code in another language...

So, it all comes to this:

1) Database: I would want to go for Neo4j. But does it goes well off when the scalability factor kicks in? (From what I could gather from google, there are no scalability issues).

2) What framework to use in case of Neo4j? I would require a framework that is able to handle tonnes of requests and large data as the users of the project would be Creating and Reading data a lot.

P.S.: I know it is a long question, but couldn't jot it down in lesser words!

Answer 1

Neo4j as a standalone server has also REST API: http://docs.neo4j.org/chunked/milestone/rest-api.html, if you can embedded your requests in single REST queries, there is no need to use native embedded neo4j. If there is no need to use the embedded neo4j, you can take any language of your choice.

Regarding the scalability, recently neo4j can be used on Azure, so it must be quite easy to scale. To learn more how to scale neo4j, go to this page on neo4j.org.

UPDATE: in the newest version of Neo4j, there is added the support for a new query language - http://blog.neo4j.org/2011/06/kiruna-stol-14-milestone-4.html.

Answer 2

I can't speak about the scalability or suitability of Neo4J for your particular project.

However, I'd strongly advise you against trying to mix and match languages like Java and PHP. It's so much easier to stick to the best one for your particular task. I'd also strongly advise you against using JNI for anything unless you have no other option. Java is fast enough that you should almost never need JNI for performance.

That said, it's OK to run Neo4j in its "full server" mode and then have your PHP or Python application access it using some driver over the network. I just wouldn't recommend making an ugly hybrid of PHP and Java at your application layer.

Some decent Java frameworks you could check out include:

• Spring
• Google Guice with Sitebricks
• Apache Struts 2

They're pretty standard in the industry and there are tons of good resources available on all of them.

Answer 3

In regards to the mini-question about language interoperability, Java provides the JNI interface, which allows the JVM and user code to make calls into other languages and vice versa. When the native code (e.g. C code called by Java, or Java called from C) runs, it is actually running in its natural environment, so there's no performance loss in terms of actual execution.