[OpenAFS-devel] dcerpc.net - freedce
Jim Rees
rees@umich.edu
Wed, 15 Aug 2001 14:04:51 -0400
You people are making me feel like an old fart.
DCE was a political reaction to what some people saw as Sun's monopoly in
the distributed computing space. This was before Windows even had TCP. DCE
was a whole bunch of unrelated technology, most of it from IBM, Apollo, DEC,
and HP, all thrown together with no overall architecture. It was a horrible
mess and deserved to die. For example, it had two separate and unrelated
directory services, for no good reason. That doesn't mean that the
individual technologies that made up DCE were no good, just that it made no
sense to put them together that way.
DCE RPC came from Apollo, where it was part of NDS. It was originally
written by Nat Mishkin and a bunch of other people, I forget exactly who,
but Joe Pato was in on it too.
At that time (about 1986) Sun was having problems because NFS had been
written to run well on Sun hardware (Motorola 68000 series) and didn't work
well on anything else, and it particularly didn't work well cross-platform.
The fact that xdr sent everything in network byte order, so that x86 had to
swap on both send and receive, is just one example of this.
NDS RPC fixed all this but it was too late, and not widely adopted.
I still have a souvenir from my DFS days:
% ls -l ~/:
lrwxr-xr-x 1 rees wheel 14 Aug 2 1994 /afs/umich.edu/user/r/e/rees/: -> /:/user/r/rees