In message <[EMAIL PROTECTED]>, Greg Skinner writes:
> The reason I am asking is because once we get to the level of hundreds
> of thousands of TLDs, we risk the DNS performance problems that have
> been discussed earlier. That would have an effect on the entire
> Internet. So it seems at the very least, we need to proceed slowly in
> adding TLDs, so we can study its effect on DNS performance and make
> changes to it if necessary to allow it to scale.
>From the domainwalk data:
+----------------------+------+-------+--------+---------+
| Country | TLD | 2LD | 3LD | Hosts |
+----------------------+------+-------+--------+---------+
| JAPAN | JP | 97 | 38979 | 1718935 |
| UNITED STATES | US | 75 | 3118 | 1642418 |
| CANADA | CA | 5048 | 259457 | 1584273 |
| GERMANY | DE | 77016 | 398631 | 1375114 |
+----------------------+------+-------+--------+---------+
Whether you do it on 2LD, 3LD or even 4LD level, it seems to work, so
why not on TLD level? It's mainly a matter of horse power, I'm quite
sure BIND is capable.
But, if one restricts applications to groups of applicants rather then
allowing individual applicants, one would avoid excessive
proliferation.
el