Springs+Cluster+FileDownload

[j2ee_chap]

I have a web application which run on single tomcat server. We use Springs MVC and DI + we have a index engine to index files like pdf, word etc.

We need to scale this application to work in a cluster now. We have our documents stored on file system (pdf, word) etc which user can download from our web interface. Also we manage indexes to be able to search. Once we go on a cluster my questions are

1. How would each node be able to serve the document which are on the file system now ? Is there a good solution other than replicating all documents manually on each server? We are hosting on windows.

2. How do we take care of indexing and searching in cluster? Ideally only one node should be allowed to write to the index and all other nodes should be able to search?

3. Do we need to move to Jboss for better cluster support?

[Petr]

We need to scale this application to work in a cluster now.

Hi
Why did you decide this solve for scalability? I guess you use Lucene for index and search data. This framework allows to use big data and you don't need use clusters.
And Spring allows to make a distributed application without use some JEE container.

[j2ee_chap]

I do not use lucene and use dtsearch as the search engine. Also the application has other features and thus scalability is not a requirement keeping on search and indexing in mind but is a general requirement for the application to server 1000+ concurrent users.

[Petr]

In this way you can use special architecture for it.
see
Actor model - Wikipedia, the free encyclopedia
SEDA - Architecture for Highly-Concurrent Server Applications