[ih] The linux router project and wifi routers

[Dave Crocker]

On 11/3/2022 6:13 PM, Guy Almes via Internet-history wrote:
>   Bosack deserves credit for understanding how big a business the 
> router business would soon be and the urgency to get it going.

A counterpoint.  This time, direct experience, rather than rumor:

While at Ungermann-Bass, around the years being discussed, I managed the 
effort to put TCP/IP onto the U-B intelligent PC card and include 
applications on the PC to use it.  Our second customer was Rutgers.  
Some months later, our salesperson called me and said that Rutgers was 
interested in buying a router from us.  I noted that we didn't have one, 
didn't have one in the pipeline, and that the salesperson already knew this.

She acknowledged all that, and further said that the Rutgers IT guy also 
knew all this.  But, she said, he liked us.  I noted that, really, he 
liked /her/.

But an opportunity is an opportunity, so I asked how long he would 
wait.  (For the router, not for her.)  She called back later and said 6 
months.

We had a guy just coming off a project who had said to me, on my second 
day at U-B, that he really wanted to built a router.  So I corralled 
Stan and asked him how long he would need to build one. We did a 
whiteboard analysis -- well, really, he did it and I just grunted 
occasionally and sometimes asked a question, mostly to keep him going.  
He estimated 5 months and I noted that was perfect, leaving us a month 
for testing.

I went to our marketing guy and he blew me off, saying that there was no 
market and only a few were needed for the backbone.  I then went to my 
boss, John Davidson (*), who said that we never like saying no.  So he 
authorized the project.

Stan Maniply was the best coder I'd seen and hit the mark.  And we 
delivered.

But U-B marketing never knew how to do anything but market to a small 
number of large customers.  (The company had a dumb Ethernet card that 
was faster and cheaper than 3-Com's, but again, had no idea how to 
market it.)

The same limitation applied to an Internet symbolic packet trace product 
I built, based on an existing U-B XNS trace tool.  At the Rutgers PC 
product delivery, I brought it with us so the IT guy could see the 
packet traffic.  He was casually satisfied with our basic delivered 
product but was got /very/ excited about the trace tool.  Lightbulb... 
This made clear that IT people doing networks would be interested in 
something that gave them insight into what the heck was going on... down 
there.  But, again, U-B marketing didn't know what to do with it.  And 
by the way, neither did Wollongong marketing, when we built a fresh tool 
there. Also, I thought it was a product opportunity, not a full business 
opportunity.  This changed when Harry Saal's company reached a million a 
month.

So, yeah, rather anti-climatically and obviously, this underscores that 
understanding an opportunity matters.  Possibly more than building a 
competent product.

d/

(*) Quite a number of people were guilty for having pointed, goaded 
and/or facilitated me down the professional path I've wandered.  John 
Davidson was an unwitting accomplice.  From his Aloha network graduate 
student venue in Hawaii, he wrote a note about delayed echoing over 
satellites, suggesting adoption of a scheme that Tenex used, allowing 
apps to offload to the operating system; this got documented as RFC 357. 
I was doing user support and had noticed echoing delays were frequent, 
more generally.  So I suggested to my office mate, Jon Postel, that this 
would make an interesting Telnet option and I'd like to write it, but 
only if Jon helped.  RFC 560, RCTE.  Option 5, as I recall.  My first 
networking technical effort.

-- 
Dave Crocker
Brandenburg InternetWorking
bbiw.net