I have a machine running Windows, where the disk has two partitions C (50 GB) and D (250GB). I do research in Information Retrieval and need to work with a large corpus (more than 50 GB) and in Linux.

So if I want to install Linux on the existing system, keeping the Windows installation intact, will it be fine to run it in a virtual box? (say, QEMU, VMWare, etc.)

An alternative is using Wubi. In that case the Linux installation has to be on drive C. Then, if I keep a small Linux installation (say 5GB) on C, and my corpus on D (mounted in Linux), how will it affect the performance of my programs which would be accessing the mounted Windows drive D.

Is it feasible to use Linux this way? Which of the above is better if at all they are a way out?

Note :

Since my post in July 2010, I have been using and have tried several ways of maintaining a disk-image that I can mount in Linux. I had a 100GB qcow2 disk and a 100GB raw disk, both formatted to an EXT3 file system.

I was mounting and connecting to the qcow2 disk using qemu-nbd. The problem was that every now and then, the connection to the disk would get lost and the running programs would throw disk I/O errors.

The raw disk would mount and work fine as a loop mounted device, but when writing data to it, the mount.ntfs program would hog the CPU and the process would take an enormous amount of time. I was in fact running make on a piece of software located on this raw disk, and after a point of time make was waiting while mount.ntfs would show 100% CPU usage.