This is still in the planning, but the looks of it, we are soon going to get a chance to build our second clustered LTSP solution. Yesterday, I had some very interesting and involved discussions with Shwetav Agarwal who volunteers at The Future Foundation School for their ICT in education initiatives.
It looks as if we are going to design and deploy at 6-node (standard COTS hardware as opposed to our other clustered LTSP solution that we are deploying using HP’s c3000 Bladesystem) clustered LTSP that designed to scale up as well as scale out. The 6-node cluster would presently comprise of standard Pentium IV boxes with 2GB RAM each. The OS is likely to be Intrepid Ibex or Fedora 10. The idea is to run the a consolidated, full-fledged thin-client infrastructure that takes care of 100+ nodes.