Figure 8.11 ASR architectural overview for Hyper-V to Hyper-V replication
The communication to ASR is via an SCVMM instance at each datacenter; you enable
this by downloading an ASR provider that installs into SCVMM and enables SCVMM
to communicate to ASR using HTTPS. A certificate is used for the secure
communications that you configure in ASR and in the SCVMM ASR provider, which
gives a mutually trusted certificate. A proxy can be configured for the communication
from SCVMM to ASR, and the only configuration is outbound HTTPS. ASR never
contacts SCVMM; all communication is initiated from SCVMM outbound to ASR.
ASR works at a cloud level. Specifically, in SCVMM at each datacenter, you need to
create one or more clouds that contain the Hyper-V host groups that will be
participating in Hyper-V Replica replication. The clouds within SCVMM are then
enabled to send information to ASR by checking the Send Configuration Data About
This Cloud To The Windows Azure Hyper-V Recovery Manager option on the General
tab of the cloud properties.
Once the clouds are known to ASR, a relationship is created between two clouds. This
is a pairing, which tells the primary cloud it has a replication relationship to a target
cloud. As part of the pairing of clouds, ASR will trigger workflows on SCVMM that
automatically configure the Hyper-V hosts for Hyper-V Replica replication using
certificate-based authentication. There is no manual configuration for Hyper-V
Replica required on the hosts. Once the cloud relationship is established, a
relationship between networks on the primary and replica cloud are configured in
ASR. This enables virtual machines to be updated with the correct connectivity when
established on the replica Hyper-V server, and a new IP address is injected into the
virtual machine from the IP pool of the new network, ensuring continued
connectivity.
A recovery plan is then created in ASR that defines groups of virtual machines to be
failed over from all of the types supported by ASR (which includes VMware and
physical hosts), defines the ordering of failover, and even defines optional scripts to be
executed that need to be present on the SCVMM servers in addition to triggering
runbooks stored in Azure Automation and even manual actions that should be