... it's just a simple reminder for people who had this question and don't find any answer.

The way to go is using a Disk Image or a Sparse Image.

There are 2 options (using DiskUtitlity):
1) Create a read/write dmg which will fit your project size -- with enough space left for some changes.
2) Create a sparseimage with a size which will allow to do a lot of changes.

Both of them behave like a local disk. If the disk is full -- no way to add files or make changes!
The sparseimage though is a bit bit different from the dmg, it can grow or shrink. This can save some disk space on the network drive for finished projects.
But again: there must be enough free space on the network drive!

Here a simple example for a dmg/sparseimage:
In DiskUtility create a new "Blank Disk Image".
Set a name both for the saved image file and the name of the disk. The entry for the "Save As" holds the name of the created file in Finder, the "Name" is the name of the disk displayed in Finder and FCPX when it's mounted.
Now enter the size with the custom settings -- for example 5G for a mini project.
Select the type: "read/write disk image" or "sparse disk image"
Save.
The dmg will cover 5G of disk space. The sparseimage about 500 MB -- as long as it's an empty sparseimage.
Both options will take a while to create the images depending on their size.
Once done they will automatically show up in FCPX and you can duplicate/save a project to those disk images.
In both cases you should use the copy of the original files and not only the references.

You even can create new events or projects on this 'virtual' disk(s). But it's not recommended from my point of view.

If you're working with Lion (or above) the image will mount automatically if the network connection is available. But you can change that in the System Preferences.

Another off topic note (or recommendation):
You should save an XML of the project, then use my MTT app to collect used FX and create a backup installer for those FX used. Though the app doesn't handle all commercial FX it will give you a survey about what's missing or not -- in case you want to use the images on another machine or a 'freshed up' machine.
MTT is free. www.spherico.de/filmtools/

-Andreas

Actually, Sparce Disk Images do not automatically shrink unless you do it at a command line.

The user manual has a good bit of info about using disk images. Steve Martin also has this guide out, too.
http://www.kenstone.net/fcp_homepage/fcp_x_managing_disk_image_martin.html

And this on OS X Sparce Disk Images:
http://en.wikipedia.org/wiki/Sparse_image

Ben you're right!

Sparse images shrink to minimum when created (means they are empty) from there they will grow.

Actually I just forgot the third option (because of the question I was asked from a user).
It was about saving a finished project to a network drive. I just duplicated my answer to that question to post it here cause this question comes up again and again. Shame on me not to list all options.

So here again:

1) In Disk Utility create a new "Blank Disk Image" (cmd-option-n).
2) Set a name both for the saved image file and the name of the disk.
The entry for the "Save As" holds the name of the created disk file in Finder
The "Name" is the name of the disk displayed in Finder and FCPX when it's mounted.
3) Now enter the size with the custom settings -- for example 5G for a mini project.
4) Select the type: "sparse bundle disk image" or "sparse disk image" or "read/write disk image"
5) Save.

Here are the behaviors:
1) "sparse bundle disk image": will create an image which is about 500 MB. It will grow up (if you add files) to the size you have entered when creating the image and will not shrink down if you delete files.
While it creates a file like under 2) it will be a segmented file (package) which looks like 1 file. It makes Time Machine working a bit faster and efficient.
2) "sparse disk image": will create am image which is about 500 MB. It will grow up (if you add files) to the size you have entered when creating the image and will not shrink down if you delete files.
Compared to 1) it's only one file.
3) "read/write disk image": will always have the size you entered when creating it.

For 1) and 2) you can use Terminal to shrink and type in hdiutil compact plus a space and drag your disk image file to the command line -- the image must be unmounted.

Working with those disk images saved on a local drive on a local machine will give you the transfer speed of the local drive, working over LAN the speed is limited to the speed of the LAN and speed of the drive where the disk image is stored. So a SAN would be speedier than a NAS in most cases.

-Andreas

SAN is faster than NAS in all cases. Nothing beats fibre. Good info here, thanks for posting.