Shulou(Shulou.com)06/02 Report--

Using the Replication replication function in Veeam Backup & Replication to copy the virtual machine from the source side to the destination side, how to make good use of the replicated backup? When there is a problem with the production system, you can perform failover failover. There is a confirmation process after the switchover. If the switchover runs normally, you can perform a permanent switchover. If the test is not normal, you can roll back.

Failover means failover, meaning that the load on the source virtual machine is transferred to the replicated virtual machine at the DR disaster recovery site.

For failure recovery, Failback switches the load on the replication virtual machine on the disaster recovery site back to the source virtual machine.

Some differences and differences in various concepts:

Replica Failover: switching the load from the production side to the replication side virtual machine is a temporary intermediate state, which can only be restored to a certain restore point of the target virtual machine. The changed data after the restore point cannot be synchronized to the replication virtual machine, which is not completely consistent with the source virtual machine.

Permannet Failover: permanent failover, that is, the status of Replica Failover is further confirmed and submitted to determine that the load of the source virtual machine is transferred to the virtual machine of the replication side, and the status of the replication virtual machine is the same as Replica Failover.

Failover Plan: for multiple dependent source virtual machines, determine the switching sequence and interval of each virtual machine, and then fail over in turn.

Planned Failover: the planned downtime and planned hardware maintenance are arranged in advance, and the load of the virtual machine on the production side is switched to the target side to replicate the virtual machine, and the data of the virtual machine is consistent with that of the source side.

Undo Failover: after the virtual machine on the production side fails over to the destination side, you can perform a permanent switchover (Permannet Failover) or cancel the failover, that is, fall back to the original state, and the workload can be switched back to the production side.

Replica Failback: switching the load of virtual machines on the replication side back to the production side means that all the virtual machines on the replication side are transferred to the production side, which is an intermediate temporary state and requires further operation.

Commit Failback: commit confirmation for Replica Failback load switching from replication-side virtual machines to production-side virtual machines

Undo Failback: when the Replica Failback load is switched from the replication virtual machine to the production virtual machine, the workload is still returned to the replication virtual machine of the disaster recovery site after the cancellation.

1. Replica Failover

Replica Failover restores the full function of the virtual machine based on a certain restore point on the target side, which can be started quickly. When performing the failover operation, the source virtual machine can still operate normally. After the necessary testing work is completed, we can also undo the failover and return to the original state. If the source virtual machine and the target virtual machine are on the same network, you can temporarily disconnect the network of the open source virtual machine to avoid IP conflicts when the source virtual machine and the target virtual machine are running at the same time.

It is recommended to use failover operation to avoid manually powering up the replica virtual machine and avoiding the loss of important data.

How to operate failover:

(1) VBR restores from snapshots in the necessary replication chain

(2) VBR starts the replica virtual machine, and the state of the replica virtual machine changes from normal to failover. If it is only used for testing, the source virtual machine is still running normally. Only when the state of replica VM changes from failover to normal, replication activies can work properly.

(3) all changes to the replication virtual machine are written to a snapshot of the change file.

Failover is an intermediate state, which requires further operation to execute undo failover,failback,permanent failover. In a disaster recovery scenario, the next step should be performed after the test is complete.

2.Permanent Failover

After performing the permanent failover, permanently switch the source virtual machine to the replication virtual machine, which replaces the source virtual machine. Suitable for source and replication virtual machines located at the same site and with similar resources, so that there is no delay in what the user is doing after the switch.

How permanent failover is handled:

(1) VBR removes the restore point of the replicated virtual machine, deletes the related files in the datastore, and all written data is submitted to the snapshot of the changed file.

(2) VBR removes VM replica from replicas list from Veeam Replicaiton console

(3) in order to protect the replication virtual machine from being destroyed after performing Permanent failover, VBR reconfigures the replication task and adds the source virtual machine to the replication exclusion list. The source virtual machine will not be replicated the next time the replication task is executed.

3. Failover Plan

When the running virtual machines are interdependent, you can use failover plan to determine the recovery sequence of each virtual machine, and set the delay interval between the started virtual machines. The default is 60 seconds, except for the last virtual machine. "you can only restore up to 10 at a time, and if more than 10, it will be queued, that is, the restore of the 11th virtual machine will not be started until the first restore is complete." Like failover, Failover Plan is an intermediate state.

Start Failover Plan

In the Failover Plans list, select a task, right, click start to restore to the latest restore point, if you want to restore to the specified restore point, you can use the start to option

4.Planned Failover

If the primary site is scheduled to be down, the workload can be switched from the primary virtual machine to the replicated virtual machine with minimal interruption. For example, when the data center is migrated, maintained and the software is in progress, Planned Failover can be performed.

Planned Failover process:

(1) the failover process triggers replication job to perform incremental replication to synchronize the unreplicated data to the replication side

(2) shutdown of the source virtual machine

(3) the failover process triggers replication job to perform another incremental copy, synchronizing the incremental changes generated at the last minute.

(4) switch the virtual machine to the replication side.

(5) the replication end starts up.

Note: this process is designed to pass the current workload to the target side, and it is not recommended to select a restore point to switch.

During the Planned failover process, VBR creates two auxiliary restore points that we can use to do rollback and other operations.

After the host of the main site is online, we can also switch back to operate similar to unplanned failover, including: undoing failover, permanent failover or failback.

5.Undo Failover

You can use undo failover to restore the replication virtual machine to its pre-failover state. VBR shuts down the replication virtual machine and reverts to the most recent snapshot in the snapshot chain. The data in the write change snapshot file is cleaned, the virtual machine is restored to the normal state, and replication activity in the source virtual machine can continue.

6. Replica Failback

The Failover operation was carried out in the previous stage, and the load of the virtual machine was switched from the production environment to the target side. After running for a period of time, if you want to switch the load from the DR disaster recovery side to the production environment, you can use Replica Failback. This operation will completely synchronize the complete data of the destination side to the production side. There are three ways:

L Failback the origin virtual machine of the origin-pull host

L Failback returns a virtual machine to a new location that was previously restored by backup

Transfer the complete file, Failback a complete new location of the virtual machine

The first two ways only need to pass the incremental part, and the third way to transfer the complete virtual machine.

The way Failback works

(1) if the source host is powered on, VBR shuts down the source host

(2) VBR creates a failback snapshot of the source host

(3) VBR calculates the difference between the switching status of the source virtual machine and the replication virtual machine

(4) VBR transfers the difference to the snapshot of the delta file of the source host.

(5) VBR shuts down the replication of virtual machines until failback is submitted or failback is revoked.

(6) VBR creates a failback protection snapshot of the replicated virtual machine

(7) VBR again calculates the difference between the source virtual machine and the replication virtual machine during the last execution of failback, and performs the synchronization operation again.

(8) VBR removes the failback snapshot on the source virtual machine and submits the Delta change file to the source virtual machine

(9) change the state of the replication virtual machine from failover to failback. VBR suspends replication activity on the source host because the switch is only an intermediate state, and the failback state needs to be committed or undone back to the failover state later.

(10) if the source host boot option is selected, VBR will boot the source host.

Failback on VSAN: due to the organization feature of VSAN data storage, VBR cannot obtain the difference between the disk of VM replica and the source virtual machine located in VSAN, so it is necessary to read all the disk data of the replication virtual machine in the difference part of the calculation, and then transfer only the difference.

Quick rollback option

(1) VBR will not query the entire disk file from the replication virtual machine, but only query CBT to get the difference.

(2) the use scenario is to copy errors at the operating system level of the virtual machine, such as application errors or mistakenly deleted files on the replicated virtual machine. Do not use the Quick rollback option if there is a hardware, power, or storage failure on the replication virtual machine

(3) it must be failback to the source location, CBT is enabled for the source virtual machine, and the "Use changed block tracking data" option is used when the replication virtual machine is created.

7. Commit Failback

Confirm the Failback operation and complete the restore to the source virtual machine. You need to submit the Commit to the Failback.

(1) VBR changes the replication virtual machine status from Failback to Normal

(2) if you are restoring to a new location, VBR reconfigures the replication task, adding the previous source virtual machine to the exclusion list, the new restored virtual machine replaces the source virtual machine in the original replication task, and when the new replication task starts, VBR will process the new restored virtual machine instead of the original source virtual machine. If it is restored to the original location, the VBR is processed in normal mode.

8. Undo Failback

When the load is transferred from the replicated virtual machine to the production environment without submitting the Commit, it is found that the virtual machine in the production environment is not working properly and can be revoked by Undo Failback.

(1) VBR deletes VBR protected snapshots

(2) VBR changes the replication virtual machine status from Failback to Failover

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.