Use availability features are the following four:
- High Availability
- Disaster Recovery
- Migration and Upgrade
- Expanding a readable copy of the one or more databases
Availability features of SQL Server can not replace the need for well-tested and reliable backup and restore strategy, which is all availability solutions essential building blocks.
AlwaysOn Availability Groups
SQL Server AlwaysOn availability group introduced in 2012 to send each transaction database to another example, to provide a level of protection database, copy the example referred to, which contains a copy of the database in a particular state.
Data movement between the replica may be synchronous or asynchronous, Enterprise version allows up to three replica sync (including primary copy).
AlwaysOn is the general term availability features in SQL Server, covering availability groups and FCI. Name AlwaysOn Availability Groups not function.
Because only provide database availability group-level protection, rather than the instance-level protection, so it is necessary for each secondary copy manually synchronize the transaction log is not captured or database that are not configured any content.
On a copy of the terms, Standard and Enterprise versions with different versions of the maximum. Standard versions of the availability group (called Basic Availability Groups) supports two copies (a primary copy and a secondary copy), and only one database availability group. Enterprise Version allows not only arranged in a plurality of database availability group, and the total number of copies may have up to nine (one primary replica, eight secondary copy).
What action to take when the behavior of synchronized copy REQUIRED_SYNCHRONIZED_SECONDARIES_TO_COMMIT. This option works as follows:
- There are three possible values: 0, 1 and 2
- Value is secondary to the number of copies that must be synchronized, its data loss, availability and failover availability group has an impact
- WSFC clusters and the case of type "None", the default value is 0, may be manually set to 1 or 2,
- In the case of a cluster type is "outside", the default value is set by the cluster mechanism, and manually override. For synchronous three copies, the default value is 1. On Linux, the value of REQUIRED SYNCHRONIZED SECONDARIES_TO_COMMIT configuration on the cluster resource availability group. On Windows, it is set by the Transact-SQL.
Value greater than 0 ensures a higher degree of data protection, because if you can not get a minor number of copies required, then until the issue is resolved, the primary copy is not available. REQUIRED_SYNCHRONIZED_SECONDARIES_TO_COMMIT also affect failover behavior, because if the appropriate number of secondary copies are not in the correct state, the fault does not occur automatically transferred. In the Linux, the value 0 if not allow automatic failover, and therefore used in conjunction with the automatic transfer synchronization failure on Linux, this value must be set to be greater than 0 in order to achieve automatic failover. On Windows Server set the value to 0 is SQL Server 2016 and earlier behavior.
SQL Server 2017 this was perfect for the following two cases provide support DTC.
- Transactions that span multiple databases in the same instance of SQL Server
- Across multiple instances of SQL Server or transactions may involve non-SQL Server data sources
The following list highlights some of the differences between the FCI on Windows Server and Linux are:
- On Windows Server, FCI part of the installation process. The FCI on Linux is in the SQL Server installation configuration.
- Linux only supports each host to install a SQL Server, so all FCI is the default instance. Windows Server supports up to 25 per WSFC has FCI.
- Linux common name in FCI definition used in DNS, and the name should be the same resources created for FCI.
Log Shipping
Log shipping transaction log backup automatically generated, and copy it to one or more instances called hot standby state, and then automatically applies the backup transaction log backup instance;
We can say that the use of log shipping to some extent biggest advantage is that it takes into account human error. Transaction logs may be delayed. So if someone issued a similar UPDATE without a WHERE clause in the case, the backup may not change, so you can switch to the standby instance when repairing the primary system.
Across data centers;
AlwaysOn Availability Groups
