Exchange 2010: How does Shadow Redundancy work?

Shadow Redundancy is another way of ensuring high availability for the Exchange 2010 Hub Transport Server. The way it works is quite simple. If a user wants to send a mail, the mail lands in a queue on a Hub Transport Server. The Hub Transport Server then establishes an SMTP connection with an Edge Transport and sends the mail to the Edge Transport Server. When the Hub Transport Server has sent the mail, it does not delete the mail from its queue but moves the mail to a shadow queue. The Edge Transport Server in turn establishes an SMTP connection with the next mail server, which could be the mail server of the recipient mailbox, for example, and marks the mail with the status "sent". After a certain time, the Hub Transport Server queries the status of the mail in the shadow queue at the Edge Transport Server. If the status of the mail is sent, the hub transport server deletes the mail from its shadow queue.

If the mail from the Edge Transport Server has not marked the mail as sent, for example because it has failed, the Hub Transport Server either resends the mail to another Edge Transport Server or sends the mail directly.

This entire process is controlled by the "Shadow Redundancy Manager". The availability of the Transport Server is determined via heartbeat. To retrieve the status of a mail (discardStatus), the Transport Server uses the SMTP command XQUERYDISCARD. The remote server then returns the discardStatus of the mails.

The configuration of ShadowRedundancy can only be done via shell. To switch Shadow Redundancy on or off, enter the following command in the Exchange Management Shell:

Set-TransportConfig -ShadowRedundancyEnabled $true

resp.

Set-TransportConfig -ShadowRedundancyEnabled $false

The following parameters are available for configuring Shadow Redundancy Heartbeat:

Set-TransportConfig -
ShadowHeartbeatRetryCount 3 –ShadowHeartbeatTimeoutInterval 300 –ShadowMessageAutoDiscardInterval 02:00:00

ShadowHeartbeatCount defines how often a server waits for a time-out before sending the message from the shadow queue again via a different route. The default value is 3 attempts.

ShdadowHeartbeatTimeOutInterval specifies the intervals at which the hearbeats occur; the default value is 300 seconds.

The ShadowMessageAutoDiscardinterval parameter defines how long a server waits before the mail is deleted from the queue, even if no server has queried the status of the mail. The default here is 2 days.