Summary the high availability technologies offered by SQL Server 2008.

来源:互联网 发布:美食杰网站seo分析 编辑:程序博客网 时间:2024/05/01 22:20

Summary the high availability technologiesoffered by SQL Server 2008.

1.      Federated Databases.

2.      Failover Clustering

3.      Mirror Database

4.      Peer to peer Replication

 ==================================================

1.Federated database is a collection ofindependent servers.Connected by a LAN.

The data is distributed across eachparticipating server.

The application see a logical single viewof data through UNION ALL views.

The application know which data in the viewis local and which is remote.

To avoid creating a hot node the DBA needto know how to partition and distribute the data.

If one node failure then need DBA to take offlinethe data on the node.

It does not provide the complete availability.

 

2.Failover Clustering is a group of SQLServer database acts just like a single SQL Server database.

The clustering does not provide additionalscalability. The database only runs on 1 node.once this node fails allconnections to this database are lost.

The new connection has to wait restart theSQL server database on a surviving nodel.

The application need to reconnect andrestart their transactions.

The procedure is complex and take rather long time and also the standby node has nothing to do if the primary node work well.

 

3.Mirror Database

The SQL Server database provide more thanone warm standby site through log shipping.

Mirror Database does not provide any kindof scalability.

The availability of it is similar to OracleData Physical Guard.

The Mirror Database is in unavailable state to applicationuntil a failover occurred automitically or manually.

 

4.Peer to Peer Replication

The Replication can move data in bothways,act as a bi directional replication.

It also provide the load balance workloadacross more than one database server.

It does not provide any scalability.Itprovide the conflict detection and conflict resolution since it isbidirectional replication.

The data is reducdent distributed across all the participating server.