>Snipped out most of your OP.
>This size of scale up would be very interesting to watch and I look
>forward to updates.
>In the past I tried doing a scale up using OpenMosix but OO does not
>survive process transfer from one server to another. Time and
>equipment constrains prevented a second trial. Some one else posted on
>the list about running 100+ clients but that was in CLI base and not
>GUI.
I have been running 29 diskless client on IBM Netfinity machine for more than 5 years and I have not seen much of the problems. This client used to run java ide like Jbuilder and all. It worked like charm. And know that same setup would be able to run another 20 machine without any problems.
>I do not think it would be NFS constrain but hardware / networking
>constrain to serve 2100 clients. If you can provide for this and
>server farm can be setp this would be intersting.
The more reading I do, I get more confused. Say GFS Vs. NFS. What kind of hardware/networking constrain you can think?
Regards
SP.
