Exchange 2007 - SMTP Connectors and cost weighting
Hi,
The flow of outbound messages are as follows:
Exchange 2007 -> MTA(IronPort) -> Internet
There are 2 MTAs and I wanted to know what happens if I set up 2 equally weighted SMTP Connectors on the Exchange 2007 server pointing to each of the MTAs. Will Exchange "load share" or round robin between the two MTAs?
What if I set up one SMTP Connector with a higher priority pointing to MTA-A and the lower priority SMTP Connector to MTA-B. If MTA-A becomes unavailable, will Exchange automatically start forwarding messages outbound via MTA-B? What if there are queued messages that were pending for MTA - A just before it went down. Will those messages just wait and Exchange will retry or is Exchange smart enough to re-queue them to go out MTA-B?
Thanks for any and all help.
The flow of outbound messages are as follows:
Exchange 2007 -> MTA(IronPort) -> Internet
There are 2 MTAs and I wanted to know what happens if I set up 2 equally weighted SMTP Connectors on the Exchange 2007 server pointing to each of the MTAs. Will Exchange "load share" or round robin between the two MTAs?
What if I set up one SMTP Connector with a higher priority pointing to MTA-A and the lower priority SMTP Connector to MTA-B. If MTA-A becomes unavailable, will Exchange automatically start forwarding messages outbound via MTA-B? What if there are queued messages that were pending for MTA - A just before it went down. Will those messages just wait and Exchange will retry or is Exchange smart enough to re-queue them to go out MTA-B?
Thanks for any and all help.
ASKER
Hi Simon,
No I have not. I will definitely take a look at it.
To your point, if there are two connectors, the one with cost 2 will not be used - even when the one with cost 1 is unavailable or down?
You mentioned that if I were to put both on the same Send Connector, they will load balance. What if one of the MTAs becomes unavailable, will it fail over to the other? What of the messages that were queued for the one MTA that became unavailable, will Exchange "re-queue" them for the other MTA that's still up?
Thanks.
No I have not. I will definitely take a look at it.
To your point, if there are two connectors, the one with cost 2 will not be used - even when the one with cost 1 is unavailable or down?
You mentioned that if I were to put both on the same Send Connector, they will load balance. What if one of the MTAs becomes unavailable, will it fail over to the other? What of the messages that were queued for the one MTA that became unavailable, will Exchange "re-queue" them for the other MTA that's still up?
Thanks.
The whole point of load balancing is that if one becomes unavailable, the other one takes over - as long as both smart hosts are on the same Send Connector.
Now if the authentication credentials required for each smart host were different, then you would use two different Send Connectors, but with the same cost - again they would load balance.
When the costs are not the same, they don't load balance. Exchange will not try another connector because the lowest one matches.
Simon.
Now if the authentication credentials required for each smart host were different, then you would use two different Send Connectors, but with the same cost - again they would load balance.
When the costs are not the same, they don't load balance. Exchange will not try another connector because the lowest one matches.
Simon.
ASKER
Can you please address fault tolerance/failover? If they are on the same Send Connector, will Exchange failover if one MTA drops off?
If they are on different Send Connectors, will Exchange failover automatically? And with different Connectors, is the requirement that they be equally weighted?
Thanks.
If they are on different Send Connectors, will Exchange failover automatically? And with different Connectors, is the requirement that they be equally weighted?
Thanks.
ASKER CERTIFIED SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
ASKER
It is clear to me now. Thanks for your patience.
On the cost basis - if you have two connectors, one with cost 1 and one with cost 2, and the address space is the same on both, then cost 2 will not be used, because the first connector is lower cost. The email will just queue.
Have you read the Technet article on Load Balancing and Fault Tolerance for Hub Transport?
Simon.