White Papers

BP1013 Best Practices for Enhancing Microsoft Exchange Server 2010 Data Protection and Availability 33
environment. The times presented in Table 7, however, show an example of how soon the
administrator can start recovering mailbox items from the recovery database.
For reference only, we show a PowerShell command that can be used to restore data from an
Exchange RDB. Where ‘RecoveryDBn’ is the name of the RDB, and ‘John Smith’ and ‘TempJSmith’ are
examples of the source and target mailboxes names (using a mixed syntax of DisplayName and Alias
attributes).
[PS]>New-MailboxRestoreRequest -SourceDatabase ‘RecoveryDBn’
-SourceStoreMailbox ‘John Smith’ -TargetMailbox ‘TempJSmith’
7.4 Restore the high availability level in a DAG
The seeding activity is required when establishing a mailbox database copy in a DAG environment.
During normal DAG operations, the passive copy of the database is kept in sync with the active copy
by the log shipping and replay processes taken care by the Exchange Replication service. It can occur
for administrative or unforeseen reasons that the passive copy is taken offline, is in a failed state, or
goes out of sync from the active copy and requires not only a regular re-sync process, but a full
update to return to normal operations. Reseeding the database after such a failure could take a long
time and consume valuable resources on Exchange hosts. To bring back the DAG to its data
redundancy status, it is important to minimize the time it takes to reseed a failed database. In this
section, we will first look at the resource consumption by the Exchange seeding processes and then
show how EqualLogic Smart Copy snapshots can help to avoid long reseeding times and resource
consumption on the Exchange hosts.
Note: For planned maintenance on passive database nodes, Exchange administrators have the
opportunity to suspend, and then resume, the replication activity. The suspension causes a side effect
where the log truncation is frozen for the entire suspension time. Because the logs can fill up the
volumes, depending on the load on the database, sometimes it can be more convenient to stop the
replication and delete the copy if the planned suspension is too long.
The initial seeding or the full update proceedings are very similar in nature. They both require a
complete transfer of the copy of the database, and generally of the Search Catalog also, unless it is
planned to have the content index generated from scratch on the passive side, which causes resource
consumption on the target Exchange host.
In the following tests, first we study the impact of DAG seeding/reseeding on the Exchange host
resources. Then we prove how to increase the level of protection of a DAG in case of a temporary
fault, with the help of the SAN infrastructure. We assessed the impact of seeding or reseeding a
mailbox database copy over the IP network using the Exchange Replication service, then we checked
the differences when executed using the iSCSI SAN network, off-loading the local host resources.
7.4.1 Testing the impact of DAG seeding over the network
We ran the tests with and without user load to identify and isolate the impact on local host resources.
We decided to show the data of the scenario without user load because it better shows the resource
utilization and the isolation of the activities was more identifiable.