Dell POWEREDGE R720XD - Manual (Page 7)

Dell High Availability and Disaster Recovery Solutions Using Microsoft SQL Server 2012 AlwaysOn
Availability Groups
7
o Asynchronous-Commit Mode: The primary replica commits the transaction without
waiting for an acknowledgement from the secondary replicas that are configured for
Asynchronous-Commit mode. This mode minimizes the transaction latency, but might
cause Asynchronous-Commit mode replicas to lag behind the primary replicas. This
may, in turn, lead to some data loss in case of failover. Up to four Asynchronous-
Commit Mode replicas (with manual failover) are supported.
 Windows Server Failover Cluster (WSFC): All availability replicas (primary and secondary)
participating in the AVG deployment must be installed on the different nodes of the same WFSC
cluster. However, it is not required that all the cluster nodes be part of AVG. Support for
Multi-Subnet clusters is introduced in SQL server 2012, and allows the SQL databases to span
long distances across datacenters.
Figure 1 shows an implementation of availability group that contains the maximum possible number of
availability replicas, i.e. one primary replica and four secondary replicas.
Availability Groups on WSFC Figure 1.
At any point in time, AVG can have only one primary replica serving read-write requests. It is possible
to configure up to four secondary replicas exclusively for read-only access.
Most common HA and DR deployments for SQL databases
Previous versions of SQL Server required that the customer combine one or more features together to
create a robust HA and DR solution for their critical SQL server databases. The most typical solutions
implemented with previous versions of SQL Server are:
 Non-shared Storage Implementations: Implementing database mirroring (with witness and
automatic failover) for High Availability, and combining it with log shipping for Disaster
Recovery. This method uses local storage (non-shared) in the primary and DR sites. This setup
can easily be replaced by AlwaysOn Availability Groups and provides both HA and DR. This
setup is a right choice for SMB customers who cannot invest more in SAN implementations.
 Shared Storage Implementations: Implementing FCI for High Availability and combining it with
Database Mirroring (Asynchronous) or Log Shipping for Disaster recovery. This method uses
shared storage in the primary site for HA and either shared or non-shared storage in the DR