Shadow Redundancy

Exchange 2010 introduces Shadow Redundancy, which expands on the Transport Dumpster (a feature first introduced in Exchange 2007) and aims to protect against all Transport server failures. Exchange 2010 Hub and Edge Transport servers retain each message in a shadow queue until each server that the message is sent to reports that the message has successfully been transferred to its next hops.

For example, if Edge Transport server A must deliver a message to Hub Transport servers B and C, which must deliver the message to Mailbox servers D and E, respectively, server A will maintain a copy of the message in its shadow queue until servers B and C report that the message was received by servers D and E. If Hub Transports B or C fail before submitting the message to servers D or E, server A will detect the failure and resubmit the message to the failover server for servers B or C. Arrows labeled (1) indicate hops where Shadow Redundancy is used.

Edge Transport servers delay acknowledgment of messages received from Internet servers until they deliver to the next hop, so if Edge Transport A fails before submitting the message to servers B or C, the Internet server will resubmit the message to server A when it doesn’t receive acknowledgment. The arrow labeled (4) indicates the hop where delayed acknowledgment is used.