On this internal interface we have the Internal Network (bound to the internal Interface)
Have to ask the guru Bob here.
My understanding is you can get the SMTP to leave via different IP via multipath rules but you can only receive on 1 IP ie the SMTP Proxy. Now that might create an issue with RDNS lookups by anti-spam servers etc
I'll check tomorrow. We have a load balanced UTM cluster 2x SG330 at one location and another 2x SG330 cluster. Each 50 miles away connected via a 1gb PtP.
Having said that, we are utilizing SMTP proxy at both sites which then relay to exchange load balancers so there are 2x smtp proxy in play here which might explain why RDNS lookups work etc.
So as below:
OUTGOING: Exchange Transport Servers (NLB) > SITE A UTM SMTP Proxy > Multipath > 50% to internet & 50% to SITE B UTM SMTP Proxy > Internet
INCOMING: Primary MX > SITE A UTM Proxy > (NLB) Exchange Transport servers
Backup MX > SITE B UTM Proxy > (NLB) Exchange Transport servers
As Louis-M mentioned, it may compromise RDNS
I use SMTP only for one exchange server, and it is in trasnparent mode (lets say im more secure this way)
I can receive emails from the second WAN (with MX record setup in DNS zone)
I am afraid to let emails going from wan2 because has different RDNS
If there isn't an RDNS entry for an IP address, your mail will most likely be bounced. We actually had this with the setup I've just mentioned above as the A & PTR record hadn't been set up. Just enabling the 2 uplink interfaces pushed 50% of the mail out through the 2nd interface and it bounced.
Fortunately, we were watching for it and expecting it also. The solution was to create the multipath rule and chuck 100% smtp via the 1st interface until we got the A, PTR & MX records in place.
As far as I'm aware, most anti-spam solutions will do a reverse lookup first to make sure the sending IP has a valid A & PTR record with it. The mail will get bounced right there if the receiving server can't connect or find a PTR record. They will then move on to the more exotic stuff like SPF, DKIM & DMARC to increase or decrease the score.
We're quite pleased with our setup and are still tweaking it as we're in the process of completely switching the sites ie SITE B (failover site) will become the primary site in the event of an outage at SITE A (primary site) becoming offline for a reason.