Switchover and Failover with Multiple Targets

In a multiple target configuration, it is important to understand how DataKeeper mirrors will work in the following scenarios:

• Manual switchover to a target server

• Source server failure followed by a manual switchover to a target server

Example:

In the following scenario, there are three servers:

• Server A (source)

• Server B (target 1)

• Server C (target 2)

Note that there are two separate mirrors and Server A is replicating to two different target volumes.

• Mirror 1: Server A → B

• Mirror 2: Server A → C

Manual Switchover to a Target Server

In the event the administrator wants to make Server B become the active (source) server, the following actions will occur:

1. Administrator initiates a switchover to Server B via the Switchover Mirror option in the DataKeeper UI.

2. Server A flushes its data to the source volume.

3. Mirror 1 is automatically deleted and recreated from Server B to Server A.

4. The mirror between Server A and Server C is also automatically deleted. (Note: There will be a few seconds delay noticed in the DataKeeper GUI; this delay can take some time based on network bandwidth and server performance.)

5. A new mirror is established between Server B and Server C. The intent log from Server A is copied to Server B. Only a partial resync of the data between Server B and Server C is required to bring them in sync. (A partial resync is the resynchronization of only the necessary data to establish the new end points and is usually much quicker than a full resync.)

RESULT

• Mirror 1: Server B → A (partial resync)

• Mirror 2: Server B → C (copy intent log from Server A, partial resync)

Source Server Failure – Manual Switchover to a Target Server

In the event the active (source) server fails, DataKeeper allows you to make Server B become the active (source) server. The following actions will occur:

1. Server A fails.

2. Administrator initiates a switchover to Server B via the “Switchover Mirror” option in the DataKeeper UI.

3. Server B deletes the local side of the mirror and creates a new mirror from Server B to Server A.

4. The mirror between Server A and Server C is deleted.

5. A new mirror is established between Server B and Server C.

6. When Server A comes back up, Server A detects that Server B became the source of the mirror while Server A was down and Server A automatically becomes the target of the mirror.

RESULT

• Mirror 1: Server B → A (partial resync when Server A comes back up)

• Mirror 2: Server B → C (partial resync)