17 Jun
2024
17 Jun
'24
10:43 p.m.
On Mon, Jun 17, 2024 at 11:56:55AM -0400, Clem Cole wrote:
What worked well was different flavors of the DFS and
the later core idea
of the VPROCS layer which I sorely miss, which allowed process migration -
which w worked well and boy did I miss later in my career. Admin of a
Locus based system was a dream because it was just one system for up to
4096 nodes in a Paragon. It also means you could migrate processes off a
node, take the node down, reboot/change and bring it back. Very cool.
After the first system was installed, adding a node was trivial, by the
way. You booted the node, "joined" the cluster, and were up.
I'm so bummed this didn't make it in the market place. I dreamed up my
own version of this, very similar, actually started BitMover to build
this but got side tracked onto BitKeeper to help Linus.
What I wanted, and sadly never got, was nodes that were small SMP
machines. Maybe 4 way. And a tricked up C that had simplistic
objects built in, locks would be automatic for the most part so
when you accessed VOP_OPEN() the lock was taken automatically.
Yes, I know that won't scale but that's fine. Scale it as far as
it can go, which is not far, and then cluster those SMP machines
to get further scaling.
The vision was to maintain a simplistic OS, not the monstrosity that
Linux has become.
For most people, a simplistic SMP OS would work fine. Then introduce
the clustering to go from 4 way to 4096 way.
I gave a bunch of talks on it, I was pushing for it in the early 1990's.
I gave the talk to some VP at SGI and they promptly hired me. Never
got anywhere while I was there but I believe they did something like
that after I left.
Woulda, coulda, shoulda.