Matthew C Aycock wrote: > This does not seem reasonable to me. I thought that share a really cheap > operation in the scheme of things. > >
A share is cheap - the particular implementation of share is not. It was designed over a decade ago when shares numbered in the tens at worst. The correct answer is the sharemgr and sharetab work which was done this year in Nevada. http://developers.sun.com/solaris/articles/nfs_zfs.html gives an overview of the work. You'll want to look at the data at the end on scaling.
