Are you sure that isn't abusing the simulation Shifty? How well did it work?
If you run the 2D Spheres demo without the barriers present, it would still work ok because the spheres never stray off course on the z axis anyway. The barriers are there to prevent the odd chance that a sphere might go off course, only then will a sphere make contact with the barriers. Whether the barriers are used or not, the physics processing time remains the same, so there is in fact no overhead in this particular example.
How you approach the problem depends on what you're trying to simulate in 2D.
If you run the 2D Spheres demo without the barriers present, it would still work ok because the spheres never stray off course on the z axis anyway. The barriers are there to prevent the odd chance that a sphere might go off course, only then will a sphere make contact with the barriers. Whether the barriers are used or not, the physics processing time remains the same, so there is in fact no overhead in this particular example.
How you approach the problem depends on what you're trying to simulate in 2D.