SQL always on and MSSRA Windows Server System Reference Architecture

Failover Clustering

Failover Clustering provides server-level redundancy on a certified Microsoft
Cluster Services configuration. The shared disk array is configured to allow all
nodes access to the disk resources with only one node actively processing.
When the server node fails, the failover cluster automatically moves the control of the shared resources to a working node and restarts SQL Server.
This configuration allows seamless failover capabilities in the event of a CPU, memory or other non-storage hardware failures.

Geographically Dispersed Failover Clustering

Geographically Dispersed Failover Clustering provides server-level redundancy on a certified Microsoft Geographically Dispersed Cluster Services configuration with one or more storage array at each site. If the site, the server node or the disks fail, the complete system and disk redundancy allows the failover cluster to handle sequent activities on the working site. This configuration removes the single-point of failure potential in a typical failover cluster configuration.

http://www.microsoft.com/sql/alwayson/overview.mspx

http://download.microsoft.com/download/d/4/c/d4cc681a-d51c-4f69-8c89-1198bee335d2/SQLServerAlwaysOnBrochure.pdf

http://www.databasejournal.com/features/mssql/article.php/3456991

These data stores do not need to be part of either the OLTP or the data warehouse environments because they are not mission critical. To prevent the data stores from affecting the performance and reliability of the other data service environments, the data stores exist on separate servers. While the data store databases do not have the same mission critical requirement as the OLTP environment, a level of availability ensures continued processing in the event of failure.

Database mirroring enables a level of availability without the hardware requirements of a separate failover cluster. The use of multiple instances has a number of benefits, one of which is the ability to make performance-tuning configuration on an
instance-by-instance basis. This enables control of the environments resource usage into the future as the data services requirement grows.

OLTP, OLAP , Data Warehouse, Data Stores, and Extranet.