We have C++ server (web applications) and Java Servers. I am not a Java developer but Java people say that Java Servers requires tomcat as it renders web pages. The Java server updates DB that C++ servers inquires upon and makes processing accordingly.

This setup is ok for a user as the user needs to have a copy of every sever for himself. Now we are in requirement for all users in our company - around 3000 people for individual development purpose. So we would need to have 3000 C++ servers for each as they will modify code for their feature work, however not all (around 20) will be for Java servers.

The issue we would then need to install 3000 Java servers (tomcat). So that each individual developer has his / her own work area and one modification will not hamper others.

Is there a way that I can reduce this setup - Number of Java app though making user independent.

So is your question, how do I set up basically 3,000 identical workstations with some particular software packages installed?

Which distro are you using? Something like FAI for Debian sounds like what you are looking for. Maybe something similar exists in the Red Hat world.

Consider Era's solution, if it doesn't satisfy your needs, you can try to automate that with expect.

I am not sure what a distro is or has functionality?

In our case we have a requirement that we need to install a suite of Servers (C++ and Java) on a Linux hosts. The issue is that they are coded to talk to a specific {IP, port} and are request based means upon request the client will find which server it needs to talk and read configured IP and port and make call furnish the request and close the connection.

Now we are in requirement of this setup for 3000 people keeping in mind that each Java servers requires Tomcat instance. So looking at the combination we have no other option but to install all the servers for each developer as they are pre-configured to talk to specific servers. The problem further is coupled that we require each Tomcat instance for each Java based server and our hardware setup is limited. The infrastructure team feels that this take up all system resource and performance would really be very very slow.

Is there any was we can re-design this approach and implement the best approachable design?

Thanks once in advance for all your help and assistance.

A Linux distribution [Wikipedia] is a collection of software built around the Linux kernel, like Slackware, Debian, Gentoo, or Red Hat. They all have different conventions and tools, although Red Hat's package management framework RPM is widely used in other Linux distributions. We assume that you are not simply downloading the latest Linux kernel from kernel.org and deploying it along with your own private, hand-picked compilation of user-space utilities (although I suppose that still happens).

So: these 3,000 Linux hosts, are they already out there, and are they running some sort of package management tool or even a well-known distribution such as Red Hat? Or are you going to install and deploy them from scratch?

Would running the Tomcat server in a virtualization environment (e.g. VMware Player) be an option?

I'm not sure I completely grasp which aspect is the main problem here; getting the servers installed and deployed, or designing a scalable infrastructure? Also the stuff about the IPs and ports, is that an installation-time problem or something which needs to continue to work throughout the servers' existence?

Well the whole idea is a pre-mature one right now and a final design is yet to be implemented for all 3000 team members.

Let me be simple as I can:

We already had a web application that was completely based on C++ servers and code. Now we have introduced new JSP pages which start a web flow and for processing they connect and pass on the information to C++ servers and process accordingly. Also with the help of some technology we are able to switch pages rendered by C++ or Java applications. The C++ and Java servers are configured to talk to particular IP and port. We have means to authenticate a valid request connection request.

In other words Java and C++ servers know whom to connect as they are configured with appropriate details.

Our project had no issues in a team of 5 people. As we modified servers and tested in our login. Now we have this setup to be pushed to live. As the project will go live the developers would need to have their own setup for their future development work (Java and C++). This is where we are facing issues.

Currently we are in strength of 3000 people and all need to have workspace for each of them in order to simulate a flow from Java servers and then interact with C++ severs for basic validations.

We have a concern that installing Tomcat and sever for so many people would greatly strain development box. This is a big issue.

The project is already over and we cannot re-design as in production its all one copy of severs.

We would like to know how we can achieve the setup for large development team.

Hence the question in concern

Looks like a hardware resources question, not software.

Try Linux cluster "Beowulf", the core engine of Google.

Beowulf is a multi-computer architecture which can be used for parallel computations. It is a system which usually consists of one server node, and one or more client nodes connected together via Ethernet or some other network

The site beowulf dot org lists many web pages about Beowulf systems built by individuals and organizations.

my 2 cents

Good luck