Develop Reference

File storage for Java based web application

I'm in the processing of designing a Java based web application (Spring based to be specific). One of the key requirement is that, this application has to accept many files of various formats (pdf, jpeg, dwg, png etc.) uploaded by the user. Also, to be able to download back to user's local computer. There will be thousands of files being uploaded/downloaded.
I am thinking of two approaches:
Upload the documents to the same box where server is running. Mostly all the documents will be uploaded to, and downloaded from box where Tomcat is running. I'm worried that, as the documents grow in number, this may impact overall performance.
Upload/download documents to another server dedicated for storing/retrieving of documents.
If 2nd approach is taken, how Spring application can upload/download files to/from remote server? Or which approach is being used in the similar applications.
Or could you suggest any other optimal way of handling this requirement.
Thanks in advance.
Ganesh

Many modern applications built like this are going to use an external storage system like Amazon S3 to store these files, which buys you all kinds of nice features - high availability for downloads, an effectively unlimited pool of disk space, data replication, and so on.
There's a tutorial available for integrating spring with Amazon S3. You should check that out. Regardless of whether you choose S3 or something else, the approach will be similar.

Have you thought about using a DB. You could store those files as BLOBs. Here is a tutorial for this: link
"This tutorial walks you through the steps of building an interesting Spring MVC web application that allows the user to upload files from her computer to the server. The files are stored in database using Hibernate."
As to two approaches you consider:
Either way you will have to manage those files, back them up and check if there is enough space to store more files. Also this may cause some security issues as you accept all files.

I recommend you using a JCR like Apache Sling or Apache Jackrabbit.
Apache Sling™ is a framework for RESTful web-applications based on an extensible content tree with the following feautures:
Content resolution that maps a request URL to a content node in the content repository
Servlet resolution that maps a content node and a request method to a Servlet handling the request
Default servlets supporting WebDAV, content creation from web forms and JSON representation
A Javascript client library, allowing access to the content repository through AJAX
Support for server-side scripting with Javascript, JSP, Ruby, Velocity and Scala
OSGi-based extensibility through Apache Felix – the Felix Web Console was originally developed by the Apache Sling project

Creating Copy of file over a network share

We have an enterprise application which saves some files on a network share (\\physicalfileserver\files\). We need to use a network share because the application is installed on multiple application server instances distributed over multiple physical server connected via LAN. Sometimes, we need to create a copy of certain files on the network share itself and save it in same the directory (on the same network share) with a different name or in a new directory (Again on the same network share). What is the most efficient way to create a copy of the file? I do not want to create any IO traffic on the network as the source and destination files reside on the same server. It also increases the time taken to create a copy of the file. This however, is difficult from What I have seen so far.
Reading files in input stream and writing them back on the server through an output stream with a different path/name.
FileUtils.copyFile() from apache commons.
NIO Channels library of JAVA (java.nio.channels.FileChannel.transferFrom())
However, all of them seem to copy the file to the server executing the program and then copy the data back to network share which isn't really required.
What can the gurus suggest to be the best possible approach to handle this situation? I can only use JAVA 1.5/1.6 due to some restrictions but am open to suggestions using higher java version as well.
The application is installed on a Windows 2008 server and the network share is also present on a Windows 2008 server.
Thanks

If you want to avoid network traffic you could write a small server in Java that would allow you to copy files locally (this would run on file server itself). Securing this would be quite important as this could pose huge security risk if somebody could use it to overwrite system files.

java application updator

I have a java application which is used many computers in my office network. This is a java swing application, and when there is an update most of the time I have to update each computer. Currently there is a mechanism to update these applications automatically but it seems like it is not working properly, that is I have a simple table in the database which hold the released versions number and when the application starting it checks its' versions against the database. If it is not tally then it downloads the whole application from the FTP server which we have installed in our network.
Recently thought to create a server to do this task. But, I don't know whether it is a good practice to do.
My idea is, there is a server which can get the md5 hashes of each file in the FTP server and send that hash list and file list to its' clients. When the clients (my swing application) get this lists it compares those hashes against its' files and if there is a mismatch client can download that file from the FTP server.
Please tell me that is this a good method to do for updating a java application?

You should use Java Webstart.
It's designed for exactly this scenario: automatic downloading and update of an application from the internet and/or intranet.
It even does clever stuff like application versioning and ensuring that new files only get downloaded when they are needed through intelligent caching.

Another option:
Since this is in an internal network. If you don't want to use webstart (best option).
Create a small front end application that use a URLClass loader and always load it on the fly. Since this is an internal network unless the application is huge. It should take less then a couple of seconds to launch.

Accessing MS Access hosted in Ubuntu server from Java in a windows based VPS?

Hi I have this little big problem I have a legacy VB6 desktop application that connects to a MS Access database hosted in a local Ubuntu server machine that is being shared to the LAN as a SMB share and I have a Tomcat web application hosted in a Windows based VPS, these are
Requirements
Read information from the MS Access db and show it in the webapp (On URL visit and almost real-time accuracy).
Update the MS Access db with information received through the website.
Facts
20Mb < db size
Shared with 15 ~ 25 users.
Constant local update and querying.
The file size is small because It is being truncated as it grows tool large( > 100Mb)
Now I know that the arquitecture isn't the best and that MS Access is at its limits, a migration to full Java and MySQL is on the way, but it's going to take a long time....in the meantime I need a way to implement that feature, here are my options.
Option 1
Access the db in ubuntu server through a VPN directly from the webapp.
Cons
Is it possible?.
Slow connection.
May lock the db MORE frequently as it happens quite often locally.
Option 2
Have a local webapp run in Ubuntu server that exposes the db as a REST API, so updates would be handled by the local webapp.
Cons:
Hard to use MS Access in a UNIX environment, looking at unixODBC and FreeTDS, but so far I haven't been able to use it.
Well, writing the whole app and securing the server.
Option 3
Any suggestions?
Thank you if you read this far, any help is really appreciated.

Unless I missed something in your description, I think you might be confusing the differences between an API and a library. Basically, ODBC is an API which is implemented as a library and commonly used on Windows based machines through additional data access libraries like ADO and ADO.NET. I mention this since you referred to unixODBC as a solution. It would not be a complete solution as there is more to it than just the API alone.
In simple terms, the database file you created with Microsoft Access is a .MDB flat file database (ok, there is a little bit more to it, but in terms of treating it as a database, that is all that matters here). If you know how the structure of the .MDB file works, you could write your own library that reads/writes to it. Of course, this is not trivial and on the Windows platforms, this is provided for you by Microsoft using the libraries included in the OS. This is also referred to as a JET driver and database. JET is the database format that the .MDB file implements and is used by Access and other applications via the correspondingly named JET drivers.
So to find an equivalent option for non-Windows platforms, you need some sort of library that knows how to natively read/write to the .MDB file directly. If you are trying to use the .MDB file at the same time from an Access application, then you need to make sure the library you choose supports simultaneous mutli-user access to the database.
In a quick search, there do appear to be some solutions that I could find. The first one appears that it might have some functional limitations. The second appears to be a commercial product.
MDB Tools
Easysoft JET/Access Driver

Please suggest some solutions for building a download service

I'm planning a web service providing file download service for handheld devices.
The requirement I have is use HTTP or FTP Protocol to build a server, use database, store files and provide Command Line Interface for devices to download requested files. No platforms and languages limit.
I'm a student and I'm learning Java Web(practicing Structs2, haven't started learnning Spring or Hibernate), and this is just a simulate project which may be required in a E-Book company. So I'm feeling like to use Java to build the service.
I'm now doing the requirement analysis, please give me some suggestions. About how to design the structure, book files stored in database or filesystem, any open source library to refer to... About any aspect is OK.
I just want to know what can I do to make it better and where I should be paying attention to.

You could implement a simple Servlet, which returns files based on request arguments.
Then, use your command line client to call the Servlet with the correct parameters, depending on what the user typed.
EDIT: more info on Java Servlet technology here. If you are learning Struts, you could implement your servlet there, but it might be a bit of overkill.