One of the best parts of my job is the opportunity to share all the cool new features that engineering has put together as a new release, so here we go! Site Recovery Manager and vSphere Replication 8.7 are now available alongside vSphere 8.0 U1, and here are the cool new features and capabilities.
SRM support for NVMeoF
Figure 1: Screenshot of storage adaptors showing NVMe-oF namespaces
NVMe over fabrics arrays continue to gain in popularity for the higher performance and higher throughput capabilities they offer over traditional SCSI or NFS devices. This is one of the more significant shifts in storage arrays in a while, and SRM 8.7 has what is required to support it.
SRM already provided support for NVMe when using vVols. With SRM 8.7, support extends to NVMe over TCP, NVMe over FC, and NVMe over RoCE. These new capabilities require changes to the SRA specification and, therefore, will require new SRAs. The NVMe options require different device identities, matching, and changes to the SRA response schema.
From a customer standpoint, once the SRA is set up, regardless of array type, using NVMe with SRM is like using SRM with NFS, iSCSI, or FC array replication. There are no changes to the UI or to functionality within SRM.
IP customization improvements
Figure 2: Screenshot showing IP customization in the recovery plan
One of the most critical capabilities of SRM is its ability to update or change the IP address of recovered VMs. Many customers rely on this feature to speed up their recoveries and to ensure that recovered applications are ready and available as quickly as possible. As security requirements have justifiably increased in recent years, this has become more of a challenge. Much of this work usually goes unseen as customers experience it as SRM continues to work as before, even though our engineers have had to do significant work to enable it.
In this release, their work is more visible as they’ve resolved a few issues that had been impacting customers' recovered VMs and recovery times.
1. Support for IP customization and in-guest custom commands for Linux VMs if the /tmp directory is mounted as noexec.
2. Start performing SRM IP customization on connected NICs attached to isolated recovery networks instead of leaving NICs disconnected during the IP customization phase. With this design change, SRM 8.7 addresses the problems with some Linux distributions and IP subnet rules and provides better RTO for Linux VMs with enabled services like networkd-wait-online.service.
Automatic VM protection updates
In SRM 8.6 we added the ability for SRM to dynamically update protection group settings based on changes in inventory mappings. SRM 8.7 will now update the protection settings for a VM when the VM is relocated to a new resource pool or folder that has an existing mapping (This capability was added for VM networks in 8.3.1). The current VM protection is preserved if there isn’t a mapping for the new folder or resource pool.
When the SRM server starts up, it will always check the current inventory mapping state of all protected VMs and update their protection if needed.
Engineering is always looking for ways of improving scale and performance. In this latest release, they have increased the number of VMs in a protection group from 500 to 1000. This allows for larger protection groups which are helpful if a customer's recovery architecture requires larger protection groups.
Another area that receives constant improvement is the user interface and how customers interact with the product. This release is no different. In this release, we’ve enhanced the datagrids that display information in multiple ways. We now support auto-fitting column data, remembering column resize settings, and allowing customers to select the number of objects they want displayed from 10-500.
New REST APIs
The SRM and vSphere Replication REST APIs were introduced in version 8.6. Engineering has continued to expand those capabilities in 8.7 by adding the following APIs:
- Create/Delete/Reconnect Pairing
- Get Protection Group/Recovery Plan Inventory
- Get Protection Group/Recovery Plan Folder Information
- Get Protection Group/Recovery Plan Folder Children
- Create/Delete/Move/Rename Protection Group Folder
- Create/Delete/Move/Rename Recovery Plan Folder
- Get All User Prompts
- Get All User Prompt Details
- Dismiss Prompt
- Get All Recovery Steps
- Get Recovery Step Details
- Create/Delete/Edit Recovery Step Callout/Prompt
- Get/Create/Delete Array Managers
- Discover/Create/Delete Replicated Array Pair
- Discover/Get Replicated Array Pair Storage Devices
- Get Storage Adaptors
- Get Storage Adaptors Connection Parameters
For the SRM appliance:
- Get All Storage Replication Adaptors
- Download/Upload/Copy/Reset SRA Configuration
- Create/Reload/Delete Storage Replication Adaptor
For all the DR appliances (SRM, VR Mgmt & VR Server), they’ve added:
• Get/Download/Generate/Delete Support Bundle
Documentation details are available here. Note that these APIs are under the new v2 path (details in the documentation)
- SRM: https://developer.vmware.com/apis/srm-rest-api/latest
- vSphere Replication: https://developer.vmware.com/apis/vsphere-replication-api/latest/
- SRM Appliance: https://developer.vmware.com/apis/srm-appliance-config-api/latest/
- vSphere Replication Management Appliance: https://developer.vmware.com/apis/vrms-appliance-config-api/latest/
- vSphere Replication Server Appliance: https://developer.vmware.com/apis/vrs-appliance-config-api/latest/
vSphere Replication has added several significant improvements as well.
Datastore cluster as a target for replication
Figure 3: Screenshot showing a selection of datastore cluster as a target for replication
vSphere Replication now supports choosing a datastore cluster as the target for replication instead of choosing a specific datastore. By utilizing storage DRS the disks are automatically placed in suitable datastores. Since this is built using storage DRS, settings like “keep all disks together” will be respected by vSphere Replication.
Exclusion of ESXi host(s) from registration
Figure 4: Screenshot showing tag assignment screen with vr disallowed host selected
Before 8.7, all hosts within a vCenter’s inventory were automatically discovered and registered to all available vSphere Replication appliances. Different behavior is preferred in some circumstances (a DMZ is one example). This new capability in vSphere Replication 8.7 allows selecting a ‘disallowed’ set of hosts that will not be registered with vSphere Replication appliances. This means they won’t be used for transferring/receiving the replication delta at the target site. These ‘disallowed’ tagged hosts can still have replicated VMs running on them. This tagging process is done using vSphere’s tagging service.
- Hosts are directly tagged as ‘disallowed.’
- Hosts can indirectly acquire ‘disallowed’ status by the containing cluster or datacenter being tagged as ‘disallowed.’
Option to retain replica base disks in the replication reconfigure wizard
Figure 5: Screenshot showing option to retain replica for excluded disks
Sometimes, customers want to remove one or more disks from replication. The current version of vSphere Replication now supports customers being able to remove the disk(s) from replication and retain those replica disks at the target site. This capability is built into the UI as a checkbox.
Figure 6: Screenshot showing a warning about reconfiguring replication without retaining replica for excluded disks
If you choose not to retain the replica of an excluded disk, vSphere Replication offers a warning before deleting it.
As you can see, this is quite the payload for a ‘minor’ release. These capabilities are in addition to all our engineering teams' work to ensure compatibility and to tweak and improve functionality that mostly goes unseen. I hope you find these features helpful, and I look forward to hearing about how you put them to good use.