Some general comments.

It’s good to see it used in at least a few places. It’s too good a system to be 
the only one using it. But I will until providence completely forces me to do 
otherwise, which I don’t anticipate. 

I’m really lucky to be able to use the system, especially in the way it was 
envisioned in the 1980’s. My first knowledge of it was when I asked Dennis 
Ritchie what was new. He said that Ken was playing around with the concept of 
union directories. Later, during one of my visits to the Labs, in 1988 I think, 
Dennis gave me a demonstration of the system.

One problem with most people who haven’t been as fortunate as I have, is they 
really just need a single system, not a distributed system. While Plan 9 makes 
a better single system for some things than most OSes, it’s really not supposed 
to have local disks at all. It really is designed to be a larger distributed 
timesharing system. At Coraid, we had two setups, one in Athens and one in 
Redwood City, that supported over 100 users in total. And without a single 
dedicated system admin. It was a very part time job, mostly for Erik, but Ian 
Ennis did some as well. It was very easy to manage because it was a single 
machine.

As far as I know, SouthSuite is now the only company both using it as a 
development system or shipping software based on Plan 9. Does anyone know of 
any others? 

Different people choose tools for different reasons and to satisfy different 
requirements the world places on them. I chose to work in embedded appliances 
so I could pick the software I use. The PIX Firewall was a bit too early for 
Plan 9—it was not yet released when I wrote the PIX—but it was very much of the 
spirit, as was the Cisco LocalDirector. Soon, we began using the 1995 Plan 9 
release and I have been using it almost exclusively ever since. I use it as the 
sole development environment and as the base of the products we ship.

In spite of our early success at Coraid with the SR, after the VC investment 
the use of Plan 9 became controversial. It’s not what others use, and in Sand 
Land (what else can one call Silicon Valley) that makes people nervous. Over my 
objections, the company attempted to move to Open Solaris. It’s a truism that a 
company that changes operating system goes out of business, and Coraid, Inc. 
again proved that to be true. The reason? A small company can’t afford the 
retooling costs to switch to another operating system. 

But things have turned out well anyway, at least for me and the traditional 
Coraid users. Now I have everything from the trademark to the source code and 
now offer the Coraid product as a software product and can support existing 
Coraid users, both with software updates and with help getting their hardware 
fixed or replaced. We are helping all those folks who bought Coraid gear 
continue to get value of their purchase. One fellow sent me a note showing that 
he’s been up over 2,000 days without rebooting. There’s never a reason to 
fork-lift an SR.

I like to think we do a good job, but our performance, efficiency and low cost 
is all made possible by the superior system that was developed by the folks at 
the Labs from 1987 thru 2002. 

  Brantley

Reply via email to