Protection and recovery of virtual machines when LUN replication is from one or more arrays on the protected site to one or more arrays on the DR site

Products

• VMware Live Recovery

Issue/Introduction

This article provides information about the limitation of virtual machines protection and recovery when the virtual machines are using virtual disk or file resources from multiple LUNs that are replicated from one or more arrays on the protected site to one or more arrays on the DR site.

This example shows a typical replication configuration that may exist in a production environment.

Example of replicated LUNs

Protected Site DR Site
-----------------------------------------
LUNsArray(s) LUNs Array(s)

-----------------------------------------

LUN-1 Array-1 LUN-1R Array-1R
LUN-2 Array-1 LUN-2R Array-1R

-----------------------------------------
LUN-3 Array-2 LUN-3R Array-2R
LUN-4 Array-2 LUN-4R Array-2R

-----------------------------------------
LUN-5 Array-1 LUN-5R Array-2R

-----------------------------------------
LUN-6 Array-2 LUN-6R Array-1R
-----------------------------------------

Replicated LUNs by array replication pairs

Array-1-->Array-1R

LUN-1------->LUN-1R - This is a VMFS datastore named DS-1
LUN-2------->LUN-2R - This is a VMFS datastore named DS-2

Array-2-->Array-2R

LUN-3------->LUN-3R - This is a VMFS datastore named DS-3
LUN-4------->LUN-4R - This is a VMFS datastore named DS-4

Array-1-->Array-2R

LUN-5------->LUN-5R- This LUN is used as an RDM LUN

Array-2-->Array-1R

LUN-6------->LUN-6R - This LUN is used as an RDM LUN

Symptoms:

Environment

VMware vCenter Site Recovery Manager 4.1.x
VMware vCenter Site Recovery Manager 1.0.x
VMware vCenter Site Recovery Manager 4.0.x
VMware vCenter Site Recovery Manager 5.5.x
VMware vCenter Site Recovery Manager 5.1.x
VMware vCenter Site Recovery Manager 5.0.x

Resolution

The protection of virtual machines in VMware vCenter Site Recovery Manager (SRM) is based on whether these machines are associated with any of the datastore groups that SRM calculates. These datastore groups may contain VMFS LUNs and/or RDM LUNs and are calculated based on what file resources the virtual machines are configured to use. These resources could be VMDK disk files, floppy disk images, CD-ROM/DVD-ROM images, or RDM LUNs. Datastore groups are also known in the SRM terminology as LUN groups.

In SRM, each and every datastore group that is calculated can only be associated with one array replication pair. That is, a datastore group can only contain LUNs from the same array replication pair.

Based on the replication example in the Purpose section of this article, SRM may calculate the following datastore groups. The actual datastore groups that SRM calculates may be less than these depending on what file resources a given virtual machine uses.

These are possible datastore groups that SRM can build:

• Datastore groups associated with array replication pair: Array-1 ---> Array-1R
• • DS-group-1: Containing LUN-1 only
  • DS-group-2: Containing LUN-2 only
    
    SRM builds these two datastore groups if both of these conditions apply:
  • • Virtual machines that use resources on LUN-1 do not use any resources from LUN-2.
    • Virtual machines that use resources on LUN-2 do not use any resources from LUN-1.
    • These virtual machines should not be using any resources from any other LUNs that are associated with any other array replication pairs.
      
      
  • DS-group-3: Containing both LUN-1 and LUN-2
    
    SRM builds this datastore group if this condition applies:
  • • One or more virtual machines use one or more resources from LUN-1 and one or more resources from LUN-2. These virtual machines should not be using any resources from any other LUNs that are associated with any other array replication pairs.
      
      
• Datastore groups associated with array pair: Array-2 ---> Array-2R
• • DS-group-4: Containing LUN-3 only
  • DS-group-5: Containing LUN-4 only
    
    SRM builds these two datastore groups if both of these conditions apply:
  • • Virtual machines that use resources on LUN-3 do not use any resources from LUN-4.
    • Virtual machines that use resources on LUN-4 do not use any resources from LUN-3.
    • These virtual machines should not be using any resources from any other LUNs that are associated with any other array replication pairs.
      
      
  • DS-group-6: Containing both LUN-3 and LUN-4
    
    SRM builds this datastore group if this condition applies:
  • • one or more virtual machines uses one or more resources from LUN-3 and one or more resources from LUN-4. These virtual machines should not be using any resources from any other LUNs that are associated with any other array replication pairs.
      
      
• Datastore groups associated with array pair: Array-1 ---> Array-2R
• • DS-group-7: Containing LUN-5
    
    SRM builds this datastore groups if this condition applies:
  • • Since LUN-5 is an RDM, the single virtual machine (or the two virtual machines in the case of MSCS cluster) that may be using this LUN should not be using any resources from any other LUNs that are associated with any other array replication pairs.
      
      
• Datastore groups associated with array pair: Array-2 ---> Array-1R
• • DS-group-8: Containing LUN-6
    
    SRM builds this datastore groups if this condition applies:
  • • Since LUN-6 is an RDM, the single virtual machine (or the two virtual machines in the case of MSCS cluster) that may be using this LUN should not be using any resources from any other LUNs that are associated with any other array replication pairs.

To know if a virtual machine can be protected SRM, all file resources used by that virtual machine must be located on LUNs that:

• Are replicated
• Belong to the same datastore group
• Are associated with the same array replication pair

If any of the three conditions are not met, SRM report that it cannot protect this virtual machine because one or more file resources it is using is not replicated. Even if the LUNs are replicated, if the other two conditions are not met, SRM still report the same error message and the virtual machine shows in SRM as Not Protected.

These scenarios illustrate where a virtual machine cannot be protected by SRM even though all file resources that are used by that virtual machine are located on or more LUNs that are all replicated.

Scenario #1

A virtual machine uses a resource from LUN-1 (or LUN-2) and another resource from LUN-3 (or LUN-4).

SRM cannot protect this virtual machine because some file resources used by this virtual machine are associated with one array pair (array pair Array-1 --> Array-1R for LUN-1 or LUN-2), while the other file resource is associated with another array replication pair (Array-2 --> Array-2R for LUN-3 or LUN-4).

In this case, SRM tries to combine LUN-1, LUN-2, LUN-3, and LUN-4 into the same datastore group but it cannot since there are two array replication pairs involved.

Scenario #2

A virtual machine uses a resource from LUN-3 (or LUN-4) and uses LUN-5 as an RDM.

SRM cannot protect this virtual machine because some file resources used by this virtual machine are associated with one array pair (array pair Array-2 --> Array-2R for LUN-3 or LUN-4), while the other file resource (LUN-5) is associated with another array replication pair (Array-1 --> Array-2R).

In this case, SRM tries to combine LUN-3, LUN-4, and LUN-5 into the same datastore group but it cannot since the there are two array replication pairs involved.

Scenario #3

A virtual machine uses a resource from LUN-1 (or LUN-2) and uses LUN-6 as an RDM.

This is similar to Scenario #2. This virtual machine cannot be protected.

Scenario #4

A virtual machine uses a resource from LUN-1 (or LUN-2), other resources from LUN-3 (or LUN-4) and uses LUN-5 as an RDM.

This is a combination of Scenario #1 and Scenario #2. This virtual machine cannot be protected.

Scenario #5

A virtual machine uses LUN-5 as an RDM and also LUN-6 as another RDM.

This virtual machine cannot be protected because LUN-5 and LUN-6 do not belong to the same datastore group and because they are associated with two separate array replication pairs.