The Internet Protocol Journal - Volume 5, Number 3

Letters to the Editor

ENUM

Ole,

I was looking at the June 2002 issue of The Internet Protocol Journal, and noticed what might be a misprint. In the story on ENUM, the next-to-last paragraph on page 21 has a sentence reading:
North America has the .164 country code of "1," implying that under ENUM there is a single DNS domain for ENUM, namely 1.e164.arpa.
I suspect it should read "... there is a single DNS domain for North America..." or something like that. (The ".164" should probably also be "E.164"—you don't refer to it as just ".164" elsewhere in the article.)

A more substantive comment on Marshall Rose's BEEP article in the same issue: It was a good overview, but I would have liked to see a mention of which application protocols are likely to use BEEP (assuming that none has already) in the near future. The middle of page 11 explains why the IETF thinks this is a good idea and why new application protocols need BEEP, but it was hard to tell whether it actually is being actively considered for use by any IETF working group.

Overall, I liked the issue, and particularly Peter Salus's review of Padlipsky's book—I came across it in the late 1980s, and actually met Michael sitting in a hallway at one of the Interop conferences before they got too big for Silicon Valley and I stopped attending. I still remember some of his cartoons and slogans (e.g., something to the effect "... the ITU is planning to have an 11-layer model because it's a sacred number in Bali..."). I've also found the articles in some of the other recent issues of the IPJ—e.g., the articles on wireless LANs (particularly the discussion of security issues) and code signing/mobile code in the March 2002 issue—very helpful, and have pointed colleagues to them.

Best wishes.

—Eric M. Berg
Managing Director, Technology Forecast Publications
PricewaterhouseCoopers Technology Centre

Eric.Berg@us.pwcglobal.com

Geoff Huston responds:

While we all try hard to eliminate various errors in manuscripts prior to publication, there are always a few author-mishaps that manage to sneak past the eagle eyes of the editor, and this is one of them.

The offending sentence should read:
North America has the .E164 country code of "1," implying that under ENUM there is a single DNS domain for ENUM in North America, namely 1.e164.arpa.
Thanks for pointing this out.

—Geoff


More about ENUM

Ole,

In the June 2002 issue of IPJ (Volume 5, Number 2), Geoff Huston wrote an interesting article about ENUM. The technical side of ENUM (using DNS to map E164 numbers to services) seems rather straightforward. But its implications on both technical and social issues are much more complex and (in my opinion) interesting. I am not an expert on the subject, but I'd like to share a few thoughts about this. First, two technical issues come to mind.

The first one is about the use of the Domain Name System (DNS). The DNS has been very successful as a distributed replicated database of hostname-to-IP address (and reverse) mappings. Will it be able to handle gracefully all the stuff people intend to put in it? This is not certain, as shown by ICANN's cautious attitude concerning the creation of new Top Level Domains. Content Distribution Networks, for example, often use lots of domain names with short TTLs, reducing the effectiveness of DNS caching (Geoff mentions this caching issue for ENUM). After all, DNS stands for "Domain Name System," not "General Purpose Infinitely Scalable Distributed Dynamic Database."

The second issue is about the status of addresses and names in the Internet. Simplifying things, we can say the following happens when somebody wants to access an Internet service with an E.164 number: The E.164 number is translated into a DNS name, and a DNS lookup gives back an URI. If the URI is a simple URL, the domain name in the URL is DNS-looked-up for an IP address, and then packets are sent to that IP address. If the URI is not a simple URL (such as a URN), some other resolving process implying the DNS occurs anyway.

That makes two levels of indirection, but, moreover, creates an "interesting" situation: IP addresses are "addresses," i.e., network-friendly identifiers, whose structure is tied to the network topology.

Such identifiers are not user friendly, so user-friendly identifiers called "names" have been created, and a "domain name system" set up to translate names into addresses. E.164 numbers are really telephone addresses. They are tied to the telephone network topology and are surely not user friendly. There are no user-friendly names in the telephone system.

The strange thing is that with ENUM, E.164 numbers are not linked anymore to the network topology, but rather become names intended for user usage. In a sense, they even are "meta names," since they translate to DNS names (that translate to addresses). But they obviously have not become user-friendly in the process.

I must admit I oversimplify a bit since I don't distinguish between names and addresses identifying level 3 (network) resources (i.e., hosts) and those identifying level 7 (application) resources (e-mails, Web pages, etc.), but this doesn't invalidate the idea.

Addresses are what the network needs, and names are what the users need. This brings me to the politics aspects of ENUM: who administers/controls/owns the namespace? A namespace is only partly technical; defining a namespace includes defining how and by whom the namespace is operated. The DNS is technically a big success, but the politics side is controversial, as shown by domain-name disputes or the setting up of alternative domain-name systems. It seems that social aspects are often more difficult to deal with than technical issues are to solve.

When I was studying networking we were taught how the technical differences between the Internet and the telephone network took their roots into a fundamental difference of culture. Now that the Internet culture seems to have won on the technical aspect (IP over broadband ISDN), wouldn't it be a strange outcome for the Internet namespace to be owned by telephone companies?

To conclude, I think this ENUM stuff shows that the Internet community really needs to work on the namespace issue, to ensure a technically and socially sound namespace for the Internet.

—Christophe Deleuze, Ph.D.
R&D Senior Engineer, ActiVia Networks

Christophe.Deleuze@ActiVia.net