4.0

How SRM Interacts with DPM and DRS During Recovery
Distributed Power Management (DPM) is a VMware facility that manages power consumption by ESX hosts.
Distributed Resource Scheduler (DRS) is a VMware facility that manages the assignment of virtual machines
to ESX hosts. When DPM and DRS are enabled on a recovery site cluster, SRM temporarily disables DPM for
the cluster and ensures that all hosts in it are powered on before recovery begins. After the recovery hosts have
been powered on, SRM relies on DRS to manage the assignment of virtual machines to hosts in the cluster.
After the recovery or test is complete, SRM re-enables DPM for the cluster, but the hosts in it are left in the
running state so that DPM can power them down as needed.
Test Bubble Networks and Datacenter Networks
To facilitate testing, SRM can create a "test bubble" network to which recovered virtual machine are connected
during a test. This network is managed by its own virtual switch, and in most cases recovered virtual machines
can use it without having to change network properties such as IP address, gateway, and so on. A datacenter
network, in contrast, is one that typically supports existing virtual machines at the recovery site. To use it,
recovered virtual machines must conform to its network address availability rules (they must use a network
address that can be served and routed by the network's switch, must be configured to use the correct gateway
and DNS host, and so on). Recovered virtual machines that use DHCP can connect to this network without
additional customization. Others require IP property customization and recovery plan steps that apply it.
Operational Limits of Site Recovery Manager
Each SRM server can support a maximum number of virtual machines, protection groups, and datastore
groups.
Table 1-2 lists the limits for a single SRM server. When you create a protection group, SRM prevents you from
exceeding the limits for protected virtual machines and protection groups. If a configuration created in an
earlier version of SRM exceeds these limits, SRM displays a warning, but allows the configuration to operate.
You should modify these configurations to bring them within the supported limits as soon as possible.
The limits for replicated datastore groups and running recovery plans are suggested and not enforced.
Table 1-2. SRM Protection Limits
Item Maximum
Protected virtual machines in total 1000
Protected virtual machines in a single protection group 500
Protection groups 150
Datastore groups 150
Simultaneously running recovery plans 3
About Failback
A failback restores the original configuration of the protected and recovery sites after a failover. You can
configure and execute a failback procedure when you are ready to restore services to the protected site.
Failback is a catch-all term for a collection of procedures that you can use to restore the original configuration
of the protected and recovery sites after a failover. The specific procedures required depend on the nature of
the preceding failover: a planned failover that leaves the protected site intact requires a different set of failback
steps than an unplanned failover initiated before (or after) an event that compromises the protected site
temporarily or permanently.
Site Recovery Manager Administration Guide
12 VMware, Inc.