dab's home page

One day down in Houston, I was driving from my hotel over to the Space Center and a little dog started chasing my car. And just because my mind works in crazy ways, I slowed down and stopped and I thought what's this dog gonna do if he catches this thing, and I watched in the rear-view mirror. The dog came stalking over very carefully, sniffed the tire, marked it territorially, turned around and walked away. And at the end of the Apollo program I thought, that is what we have done; we caught the moon, we peed on it, and we left.

- Bill Hines, Chicago Sun-Times

Pictures

Horton Trip
In the summer of 2003, I went on a 23 day canoe trip down the Horton River in the Northwest Territories of Canada.
Iceland trip
A Smithsonian Journey's trip to Iceland in June 2006. I haven't edited these pictures yet to remove the crappy and redundant ones so just overlook them.
Alaska - May 2007
A friend and I went up to Alaska for the Alaska Airmen's Tradeshow in Anchorage and then the Valdez May Day Fly-In.
Assorted Pictures
Assorted pictures that I've put on the server.

Aviation

Stallion
I'm building this airplane but my web page is very out of date.  Progress on the plane has continued however and we'll get there.

Work

History:

Pirate Airworks
Flying Beaver floatplanes in Ketchikan, Alaska
Kantishna Air Taxi
Flying Cessna 206s in Denali National Park, Alaska.
Green Flash Networks
A small ISP delivering wirelss Internet service to marinas and anchorages in St. Thomas in the US Virgin Islands.
InternetShare
We wrote a neat little piece of router technology that looked like a kind of NAT but could also translate between IPv4 and IPv6 and dynamically build tunnels to send v6 packets over the v4 Internet. The idea was that IPv4-only hosts behind this NAT on each end, could communicate with each other with no modification of the host needed. Since we used IPv6 to achieve this, it would also work as a v4 to v6 transistion tool.
ISI
Embedded Operating Systems and development tools.
Epilogue Technologies
Embedded network protocol implementations: IP, TCP, SNMP, and others.
FTP Software
IP implementation for MS-DOS (remember MS-DOS?).
MIT Telecommunications
Deployed one of the first multi-protocol routers.

Writing

Assorted writing from my time in Internet protocol design.

New Network Layer for the Internet
Back in the early 1990's, people involved with the Internet noticed there were a couple major problems on the horizon: we'd soon be running out of addresses and the routing tables in the core of the net were rapidly filling up. There were also some less immediate but long standing problems that seemed like good things to fix while we were at it (multihoming and separation of identifier and address for instance). A quick patch was instituted for the routing table problem (CIDR) and work began to find a real fix. Eventually this work lead to the adoption of IPv6 which gave us more address space but, unfortunately, did not address any of the other issues.
Here were my thoughts about what a replacement for IP requires; this is basically the thinking that leads to a Nimrod type architecture.  I don't think this was written very well and it was largely a rehash of stuff Noel had written, but I was straightening out my own thoughts on the issue.
Addressing in Nimrod
As we were working through issues with Nimrod, some of us kept coming up against the question of top-down or bottom-up locators (addresses).  Noel had written about doing it bottom-up, but almost all our designs ended up top-down.  Since I really liked the idea of bottom-up addressing, I thought about it until I came up with this idea and, in a manner of speaking, got rid of addresses entirely.
Proxy instead of sub-agents in SNMP
In the early days of SNMP, sub-agents were a hot topic: how to implement them correctly with respect to SNMP, should we modify SNMP to make sub-agents easier, is this a protocol or a programming interface, should we standardize sub-agents, was the IETF the right place, and so forth.  I came to the conclusion that the best solution was to skip sub-agents entirely, use the proxy capability that was already in SNMP, perhaps augmented by a proxy MIB, and write better SNMP management stations.  When the Simple Times did an issue on the topic, I weighed in with my proxy argument.  In the end, I lost and the IETF did a sub-agent protocol.

Misc

Links
This is what I use for my home page.

David Bridgham

Last modified: Tue Sep 26 15:43:28 EDT 2006