[p2p-research] Google gets into the DNS business

M. Fioretti mfioretti at nexaima.net
Sun Dec 6 17:19:56 CET 2009


On Sun, Dec 06, 2009 16:19:36 PM +0100, Eugen Leitl wrote:
> 
> > servers. Port 25 incoming (=receiving email from the Internet for
> 
> Sharing services can be typically on arbitrary ports.

Eugen,

to the best of my knowledge, you cannot specify port numbers in mx DNS
records. See "incoming SMTP port" at
http://www.code-crafters.com/abilitymailserver1/smtp.html

This means that if you run the email server at home but your
ISP blocks port 25 incoming, other mail servers will find yours (by
reading the mx record) but won't be able to talk with it: because they
will try to do it through port 25 which is blocked by the ISP, and
won't know which other port they could use.

> > Besides, any home computer is much less reliable than a server in a
> > professional datacenter. If a fuse blows up at home and fries your
> > computer while you're on a business trip, you're toast. And I don't
>
> If your service provider eats your data you've got not even a
> bricked piece of hardware to show for it.

All the mailboxes and configuration of my VPS mail server are already
regularly backed up elsewhere. The probability of blackouts, hardware
failures, hardware theft, unavailable connectivity and so on are
orders of magnitude lower in a data center than in any private
residence, so I'll continue to use a remote VPS. If they screw up, I
can find another provider, get a new account and migrate there all my
data and email management in two days. Communication privacy would
really exist only if it were possible to only use crypted email for
all personal messages, but that is still impossible because of people,
so it wasn't an issue when I decided how to manage my email.

This said, a couple of comments:

I'm talking mostly about communications, you mostly about access to
one's private data. They're both important, but different problems.

Personally I would have no problem in continuing discussions about
NAT, dynamic DNS and so on, but there's no better way I could find to
make sure that nobody else on the list participates. I'd rather avoid
deeply technical discussions here.

Marco



More information about the p2presearch mailing list