Well the stuff that I am running in the vmaware image roughly needs around 10GB of RAM and when I run the image which requires 8 - 10GBs of RAM on windows vista with 2GBs of RAM, it takes for ever to open up files, make code changes, build & deploy, restarting servers etc could take upto an hour.
Sometimes the change is as minor as putting a single line statement in a java file but from writing one line of java code to actually testing it could take 50 - 55minutes easiy.

Now I noticed that Fedora 8 64 bit was taking 1300MBs of RAM and leaving only 1700MB for other applicatons (See my first post). This is worst than windows Vista.
Is it possible to limit Linux OS memory usage to 128MB, 256MB or 384MB?
That way I can allocated more than 2GBs to the VMWare image?

The 1300MB is misleading. It is just preloading as much as it thinks useful. If you need more than the remaining 1700, it will release part of the 1300 without any issues.

By the way, you need to go into BIOS and enable memory mapping to see more than 3GB.

My god, what is that, 8 to 10 GB?? How are you going to use that much if you have only 4 installed? And, by the way, certain VMWare products support only 2GB and 2 cores per image.

Trying to run something this big on a laptop is just beating your head against the wall. Set up a decent machine to do this kind of thing. If you are regularly running 8-10gb files get a machine setup with 12GB+ of ram. You will be shocked at how much faster things will go. Going with a big machine also lets you get into much more powerful CPUs for very little money(relative to a laptop). If you need the mobility just VNC into the big machine from your laptop.

Jay is right about the way Linux handles memory.

Well BIOS seems to be very restrictive on this particular machine. I see options such as AGP Aperture Size on my other machines' BIOS Setup, but do not see that option on this machine and alot of other things seems to be missing. They probably are there but atleast I don't see them on the BIOS Setup Screen. I read about memory mapping but I don't see any option to enable memory mapping on this machine by going into BIOS Setup.

So what is on the VMWare Image?
- DBMS (Not sure how memory it needs)
- CMS ( needs 2GBs of RAM minimum )
- 2 Web Applications ( need atleast 1GB of RAM for each)
- BEA Weblogic Server ( I think 512MB is dedicated to it )
- IDE ( Easily take upto 512MB of RAM )
- JVM ( Need to allocated upto 512MB of RAM to this, otherwise Java.lang.OutofMemory errors are thrown )
- There are many more small players or pieces in this puzzle which needs to be up and running at all times. All the small pieces take anywhere from 1.5GBs to 2GBs of memory.

This is obviously not the right of way doing things but all the above mentioned pieces are dependent upon each other and it impossible to setup a distributed environment for 10 developers and sharing one instance is not possible since developers would start getting into each others' way.

Fedora may "seem" to be taking up 1.7GiB of RAM, but, that's a good thing. Unlike Vista that takes ~800MB alone for the OS to idle, Linux performs HDD cacheing to minimize the time it takes to load files and programs. Take a look at my RAM usage (Fedora 8 x64, 2GB RAM total)

So, it says I'm using 1854MB RAM. 990 of that is HDD Cache. So I'm actually only using 864MB RAM. I have a lot of stuff running ATM, KDE, Compiz, Firefox, OpenOffice, Amarok, and a RedHat 7.2 Virtual Machine running(512MB allowed). If I were to open let's say a Solaris VM that needs 128MB RAM, it would drop cached space for RAM. And I have an untouched swap partition.

If you need more RAM in your VM then tell VMware to allow it. Edit Vitural Machine -> memory Slider. I am using VMware Server, it allows me to allocate up to 3508MB of RAM to my Red Hat VM.

Armanox, I am not saying VMWare is not allowing it, VMWare allows me to allocate as much RAM as I want. 32 Bit Windows 0S doesn't allow any one process to reserve more than 2GBs of RAM (Windows issue, not VMWare's).
Anyway so here is the final question:
- Because of Intel's 32 bit Chipset, no OS on my machine will ever recognize more than 3072MB of RAM, even though I have got 4096MB of RAM instaled in my laptop
- Windows Vista lets me allocate upto 2.2GB (at most) to my VMWare Image.
- Vista is trash, forget about it
- Linux is the way to go
- 32 Bit Linux or 64 Bit Linux? Both of them will only see 3GBs of RAM on my Laptop but which one would you guys recommend running?
- Based on the answers from all the linux experts, it sounds like running a VMWare image with 2.2GB RAM allocated to it would be a much better experience on Linux than on Windows VISTA? Is this conclusion correct?
Ok so 32 bit linux or 64 bit linux? or it doesn't matter? Because with Windows VISTA, I have heard that 64 bit vista is more resource intensive than 32 bit vista! is that the case with Linux also?

Thanks guys.

Ok, let me give this a stab.

I personally recomend 64bit Linux. On my HP Laptop, a Turion64x2, I originally installed Fedora for IA-32 thinking about Flash issues and driver support. I lacked sound, wifi, and the system was sluggish. When I reformated the partition, I installed Fedora for x64. The system used slightly more resources, but, ran 10x as fast. Not only that, but suddenly all of my drivers worked correctly! So yes, x64 is more resource intensive then IA-32, but, it feels like it's using less. So I would recommend using 64bit, and I would expect it too be a much better VM experience then Vista.

Depends. I remember creating an XP64 image some time ago. 2 cores and 2GB of RAM and it came out slow as hell - until I did some research. Apparently, over-assigning RAM can be counterproductive. I noticed a tremendous performance boost by lowering RAM to 1.2GB (How intuitive is that?). Disabling Windows virtual memory was also very helpful (you simply don't want a virtual machine to start swapping - this may be the chief cause of the slowness you experienced). You may also need to edit your memory settings and disable memory sharing and trimming. And installing vmware tools is strongly recommended.

But one question I have is this: why would you run all those applications in a virtual machine? Wouldn't it make far more sense to develop them on your native system and run them inside the image only for testing purposes?

I also think your estimate of 8-10GB is way off. You wouldn't simply add up all the requirements. If something requires, say, 2GB that is really a worst-case scenario. On top of that, it doesn't imply that you should have 2GB free, you should have 2GB to be able to run that application without crippling the rest of your system.

Before I can test anything, I need to build and deploy my code also. Please read below to know the time it takes to test the code I have written. Testing my code has two parts
1- Build & Deploy
2- Restart Server(s)

My Estimates:
I agree with you to some extent about my estimates, my estimates are off but still all of the pieces I mentioned in an earlier post of mine, are supposed to be installed and run on a dedicated server (machine). Whereas I am putting everything in one virtual machine. Obviously this is my personal development environment and the load on any of the pieces I mentioned is next to nothing. But when I write up some code and do a "Build & Deploy", it takes about 15-18 minutes to do a build and deploy. Compare 15-18 minutes of code deployment on my vmware image with less than a minute of "Build & Deploy" of the same code on an actual distributed environment (not a virtual environment).
Bringing up BEA Weblogic Server could take anywhere from 25 - 30 minutes on my virtual machine. I don't think the server which is dedicated to run Weblogic would take more than 2 minutes to be up and running.