Sad Clouds via Postfix-users:
> Hello, I've been thinking about different ways of load balancing
> between two SMTP servers via DNS round-robin. I can see at least two
> different methods.
>
> Method 1 - Single MX record that points to a host that resolves to
> multiple IP addresses:
>
> # Individual SMTP servers
> local-data: "smtp1.example.com. IN A 10.0.5.51"
> local-data: "smtp2.example.com. IN A 10.0.5.52"
This works because the Postfix SMTP client randomizes the order of
equal-preference IP addresses.
> # DNS round-robin for smtp.example.com via multiple A records
> local-data: "smtp.example.com. IN A 10.0.5.51"
> local-data: "smtp.example.com. IN A 10.0.5.52"
This works because the Postfix SMTP client randomizes the order of
equal-preference IP addresses.
> # Single MX record that resolves to multiple IP addresses
> local-data: "example.com. MX 10 smtp.example.com."
This works because the Postfix SMTP client randomizes the order of
equal-preference IP addresses.
> Method 2 - Multiple MX records with equal priority that point to hosts
> that each resolve to a single IP address:
>
> # Individual SMTP servers
> local-data: "smtp1.example.com. IN A 10.0.5.51"
> local-data: "smtp2.example.com. IN A 10.0.5.52"
This works because the Postfix SMTP client randomizes the order of
equal-preference IP addresses.
> # Multiple MX records
> local-data: "example.com. MX 10 smtp1.example.com."
> local-data: "example.com. MX 10 smtp2.example.com."
This works because the Postfix SMTP client randomizes the order of
equal-preference IP addresses.
> I think method No 2 is the traditional way of specifying multiple SMTP
> servers in DNS. Can anyone suggest specific advantages/disadvantages
> for both methods, or are they functionally the same?
They provide the same result for the reasons stated above. The name
in the A record matters only for TLS.
Wietse
_______________________________________________
Postfix-users mailing list -- [email protected]
To unsubscribe send an email to [email protected]
