When we discover we didn't get it quite right....

In the cyber-physical systems group we had a discussion about whether there should be a requirement that it should be possible to escape from the architecture. This was described as the 'non-bad NAT'. I.e., a controlled way to violate the architecture when it is discovered that it doesn't meet a set of needs. The motivation for this was to support new physical systems which have special needs that may not be met by the architecture.

I'm not sure whether this is a good idea or even possible but thought it might be worth a little discussion.

Views: 49

Reply to This

Replies to This Discussion

In addition to "not getting it right," there issue of longevity and technical evolution at play. Cyber-physical systems have an interesting (somewhat unique?) property that the technology (processors, sensors, etc.) are evolving very quickly, yet once deployed may be in place for decades. As noted in the discussion yesterday, cyber-physical networks would not typically need to be fully integrated into “The Internet,” but rather should be connected through gateways of various types. I think that the idea of the architecture of “The Future Internet” explicitly accommodating such gateways rather than treating them as “evil” as in today’s NAT devices and similar gateways is, indeed, an interesting idea.
Wow. NAT is not evil at all. It merely uses a local network as a virtual host on a larger network.

On the other hand, the need to perform NAT due to shortage of IPv4 numbers is evil, and should be corrected as soon as possible by expansion to IPv6 addresses.
I should have put "evil" in quotes as that was the word used in yesterday's discussion. It is "evil" in its violation of end-to-end semantics (and its exposing of poorly designed applications, but we should blame NAT for that). Of course, it is definitely not evil in allowing me to have my own private network that can still participate in the global Internet.
OK, I'm getting into quibble mode. But I think that quibbles help shake out terminology as long as we don't spend too much time/energy on them.

NAT doesn't violate "end-to-end semantics" at all. It merely creates a "virtual" endpoint out of a local network.

On the other hand, I think that almost all current uses of NAT are undesirable, done for mistaken reasons when we didn't really want such a virtual endpoint.

Scott Midkiff said:
I should have put "evil" in quotes as that was the word used in yesterday's discussion. It is "evil" in its violation of end-to-end semantics (and its exposing of poorly designed applications, but we should blame NAT for that). Of course, it is definitely not evil in allowing me to have my own private network that can still participate in the global Internet.

Reply to Discussion

RSS

© 2024   Created by David Clark.   Powered by

Report an Issue  |  Terms of Service