I have been screwing around with e-mail based product development work the last
 5 years and so I have a fairly good grasp of how e-mails flow along
the Internet.

Now that reminds me long ago in college I answered a paper about how
e-mail flows
 on the net. I got it wrong. I thought that just like packets flow
from hop to hop mails
 are stored at each hop.

I started with spam control and now I also have a mail server.
Basically an e-mail
 that is sent from point A to reach point B on the Internet has to
pass through multiple
 routers the same way a web server is reached.

But there are lot of differences. The primary concept is that of how
DNS plays with it.

A DNS record called the MX pointer is key here.

For you to receive mail from the outside world a DNS record has to
tell them that
the domain you are hosting can be reached on such and such IP address.

An MX record is not an IP address. It is a mapping between a domain
name and another
 domain name whose A pointer points to the IP address.

$ dig zoho.com mx

will tell you. Or

$ dig +short gmail.com mx

Anyway the point is that for a mail server to function properly it
should both send and receive mail traffic.

A simple mail server is easily configured using any worthy UNIX
operating system. It comes with stock
 sendmail, postfix, qmail, exim or whatever else.

The popular solutions like Zarafa, Zimbra, Scalix, Kerio and so on are
built on top of Postfix or sendmail or
 any such.

I use postfix.

It is far easier to be able to send mail out than to receive them as
sending does not involve MX pointers.

Only static IP that is not blacklisted is necessary.

If you have that you can easily send mail that is received by most servers.

In that regard you may check the popular blacklist sites like
senderbase.org used by Ironport which hurt
 my customer a lot these days(I somehow fixed the issue) and
blacklistcheck.org or some such thing.

If you have a DHCP address like in a normal broadband then dont run a
mailserver. You can only when
 you can authenticate against gmail and wish to use a smarthost to
transmit and receive mail.

Not much fun in running such a mail server.

Now for how spam control works.

I guess most of the technically minded clued in guys would have
followed something about e-mail servers.

Spam control falls into a technology area identified by Gartner and
other MBA companies as
 e-mail security gateway solutions or network security in case you
talk UTM as I wrote two days ago.

It is about controlling unwanted mail at the gateway/network level.

But just like virus scanning this can be done at the user mailbox
level as well using simple Outlook rules.

But almost in every case spam control is performed at the mail server.

We will discuss that later.

E-mails are sent and received using port 25 of TCP, port 587 called as
submission port is used in certain
cases where it is normally authenticated.

And port 110 and 143 are used for POP3 and IMAP4 for downloading mails
from mail server to mail clients
 like Outlook also called as MUA. mutt is the MUA I use in UNIX.

Dovecot is the best server for this. It handles authentication for
POP3 and IMAP and serves mail from mailboxes.

There are other secure variants that use SSL and SMTP auth is another
topic for mail relay authentication.

But for sure authentication is done during mail download to identify
the user's mailbox.

For mail relaying/sending it is not a must as the authentication is
performed by the IP address as it belongs to the
 accepted network of Postfix.

This is where a VPN is very helpful as you can belong to the local
network segment from anywhere in the world
 by using OpenVPN and connecting to the company's mail server.

If not you can use the submission port 587 and do SMTP auth to send mail.

I have not configured this so far.

-Girish


-- 
Gayatri Hitech
http://gayatri-hitech.com
_______________________________________________
ILUGC Mailing List:
http://www.ae.iitm.ac.in/mailman/listinfo/ilugc

Reply via email to