Database Reference
In-Depth Information
The automatic page repair process on the principal database is attempted when the node is in
a SYNCHRONIZED state.
Automatic page repair is the enhancement to the database mirroring feature from SQL Server
2008 onwards that will perform a repair process and replace it with the corresponding copy
of its partner. The sequence to attempt automatic page repair is, when a data page error
occurs, the principal server inserts a row in the suspect_pages table in the msdb system
database. Briefly, the page is marked as restore pending, which will be inaccessible for normal
transactions and the same page will be replaced from the mirror database.
The page types such as File header, Database boot, and
Allocation pages cannot be automatically repaired.
The log stream compression is applied automatically irrespective of operative mode that
is chosen during database mirroring setup. Before sending the log records to the MIRROR
database, the stream is compressed to reduce the network latency, which improves the
performance of a synchronous mirroring topology.
The log performance improvements result in a write-ahead on the incoming log stream
on the mirror server that writes the incoming log records to the disk asynchronously. This
improvement uses log-send buffers that are used at the undo-phase, after a failover to
improve the speed of the undo phase.
See also
To configure the database mirroring using wizard, refer to the Implementing Availability
feature enhancements recipe mentioned in Chapter 2 .
Managing Database Mirroring in a Failover
Cluster environment
The two important high-availability features from SQL Server 2008 R2 are failover clustering
and database mirroring. Both of these features are fundamentally different in terms of
implementation, having instance level availability for failover clustering and database level
availability for database mirroring.
In terms of architecture, both have the same concept of quorum, heartbeat, and witness
server to support automatic (for database mirroring it depends on the implementation mode)
and manual failover. In terms of connectivity, during a failure, failover clustering does not
require any changes in the application connection string. However, for database mirroring,
the application may require changes to the connection string, to force a reconnection after
a failover. In such cases, ensure to cache the partner information to update the connection
string, or retry connectivity when the first connection attempt fails.
 
Search WWH ::




Custom Search