SRM Protection Group Overview
Site Recovery Manager offers a grouping mechanism for related datastores and/or RDMs to be failed over together called a Protection Group. SRM protection groups can be one of three types:
- Datastore Groups. This is a related group of datastores or RDMs. Currently the FlashArray does not enforce that any particular volumes be failed over together. Therefore, datastores and RDMs are only required to be failed over together if there is one or more virtual machines that span multiple volumes (e.g. a VM that has a virtual disk on a VMFS and also uses an RDM--the volume underpinning the VMFS and the volume underpinning the RDM will be grouped by SRM).
- Individual VMs. This is not relevant when using FlashArray replication and is only for VMs protected by vSphere Replication. While it is supported to use vSphere Replication on virtual machines hosted on Pure Storage, SRM does not support protecting VMs in SRM with both options. So the best practice is to use either vSphere Replication or FlashArray replication, not both.
- Storage Policies. This is a grouping of replicated datastores via storage policies. This provides for the ability to move VMs between datastores without breaking SRM protection.
Replicated volumes are added to a protection group via one (and only one) of the above mechanisms. In order for a volume to be added to a protection group, they must be discovered as replicated from a replication array pair. Furthermore if they are to be protected by the stretched storage feature of SRM, they must also be protected with a storage policy.
An SRM protection group is then included in one or more SRM recovery plans. A protection group defines what datastores/RDM are related, which then in turn dictates what virtual machines are related. All VMs that use a datastore or RDM in a protection group must be failed over together. A recovery plan then describes how a VM is to be recovered (should it be powered-in, what order should it power-on in, should a script be run, an IP be changed, etc.). To have the ability to fail over the same VMs in different ways, add the protection group that protects that virtual machine to different recovery plans.
Creating an SRM Protection Group
To create a protection group, click the Protection Group link at the top of the SRM web interface:
Then click New.
Give the protection group a name and optionally a description. Choose the direction for the protection group. A protection group can only protect datastore/RDMs that are being replicated in the same direction. In other words, a protection group cannot have datastores/RDMs being used in vCenter A and replicated to vCenter B as well as datastores/RDMs that are in use in vCenter B being replicated back to vCenter A. Only one direction or the other is supported at once.
Now choose the type of protection group.
Datastore Group Based Protection Groups
Once selected, available array pairs will be listed. Choose an array pair that is enabled (green checkmark). If the array pair is not listed, ensure the array manager is configured correctly and/or the arrays have a replication connection between them.
The wizard will then list all of the valid replication devices for that array pair. If the devices are grouped it means that one or more VMs are using multiple devices at once, therefore SRM is enforcing their grouping.
You can select one or many groups (and you can add or remove more later as needed) to add them to this protection group.
For ActiveDR, consistency groups are advertised to SRM and that consistency group includes all devices in an ActiveDR source pod. This will then force all volumes in an ActiveDR pod to be in the same datastore group and therefore they are all required to be managed by the same SRM protection group.
Storage Policy Based Protection Groups
An alternative to using datastore groups for protection groups is tag-based storage policies. Storage policies allow for greater flexibility via a few ways:
- Virtual machines can be migrated between datastores manually or automatically and storage policies will make sure they remain on the correct datastores
- If a new replicated datastore is provisioned and new VMs put on it (or pre-existing ones are moved to it) the datastore will automatically be added to the SRM protection group for the corresponding policy
- Datastores for different array pairs can be added to the same SRM protection group
- Only method supported for stretched storage-protected VMs
Note that VMs that are on ActiveCluster volumes can be protected in two ways; either via the stretched storage feature of SRM (long distance vMotion of the VMs between arrays, or disaster restart) that keeps the volumes on the stretched volumes, or the via asynchronous periodic replication from an ActiveCluster pod to a third remote array. The former method must be configured with storage policy-based protection groups. The latter method (failover to third array) can be protected by either a storage policy-based protection group or a traditional datastore group-based protection group.
The main limitation around storage policy-based protection groups is that they do not support virtual machines with Raw Device Mappings. For VMs with RDMs, you must use datastore group-based protection groups.
Note that for VMs and datastores to be protected via storage policies you must configure tag categories, tags, assign the tags, create the policies, and assign policies to the VMs on the datastores. See the below article for more detail:
To add a storage policy to a protection group, choose Storage Policies
Then choose one or more storage policies to add to the protection group. This should be a tag-based storage policy--SRM will not filter out non-applicable policies.
vVol Storage Policy Based Protection Groups
Starting with Site Recovery Manager 8.3, you can use add vVol-based virtual machines protected by FlashArray periodic replication to SRM protection groups.
To see the configuration steps of vVols and vVol replication, refer to this article:
To create a vVol-based protection group, login to SRM and click on Protection Groups, then New:
Give the protection group a name and description, then click Next.
Then choose Virtual Volumes (vVol replication). Below SRM will then list all discovered replication sources (in this case a FlashArray) for the previously selected source vCenter. Choose one and click Next.
The next screen will show all available replication groups and the VMs assigned to them. Choose one or more. You can filter out empty groups by clicking the filter slider at the bottom:
Do not add a replication group that has more than one FlashArray replication targets. VMware SRM does not support replication groups that replicate to more than one target at this time.
Complete the process by adding it to an existing recovery plan or a new one.
SRM Protection Groups vs. a FlashArray Protection Groups
People who are familiar with Pure Storage FlashArray terms, would recognize this name--the FlashArray also uses the term Protection Groups. It is important to note that these are related mechanisms (in that they both have to do with replication) but they are not the same.
An SRM protection group includes one or more volumes that must be failed over together. A volume can only be in one SRM protection group at a time.
A FlashArray protection group is a replication and/or local snapshot policy that can be assigned to one or more volumes on a FlashArray. Many volumes can be assigned in a FlashArray protection group and a volume can be in many FlashArray protection groups.