This post will walk through the installation of Site Recovery Manager (SRM) to protect virtual machines from site failure. SRM plugs into vCenter to protect virtual machines replicated to a failover site using array based replication or vSphere replication. In the event of a site outage, or outage of components within a site meaning production virtual machines can no longer run there; SRM brings online the replicated datastore and VMs in vSphere, with a whole bunch of automated customisation options such as assigning new IP addresses, boot orders, dependencies, running scripts, etc. After a failover SRM can reverse the replication direction and protect virtual machines ready to fail back, all from within the vSphere web client.
- SRM is installed on a Windows machine at the protected site and the recovery site. SRM requires an absolute minimum of 2vCPU, 2 GB RAM and 5 GB disk available, more is recommended for large environments and installations with an embedded database.
- The Windows server should have User Access Control (UAC) disabled (in the registry, not just set to never notify) as this interferes with the install.
- Each SRM installation requires its own database, this can be embedded for small deployments, or external for large deployments.
- A vCenter Server must be in place at both the protected site and the recovery site.
- SRM supports both embedded and external Platform Services Controller deployments. If the external deployment method is used ensure the vCenter at the failover site is able to connect to the Platform Services Controller (i.e. it isn’t in the primary site). For more information click here.
- The vCenter Server, Platform Services Controller, and SRM versions must be the same on both sites.
- You will need the credentials of the vCenter Server SSO administrator for both sites.
- For vCenter Server 6.0 U2 compatibility use SRM v6.1.1, vCenter Server 6.0 U3 use SRM v6.1.2 and for vCenter Server 6.5 and 6.5 U1 use v6.5 or v6.5.1 of SRM.
- Check compatibility of other VMware products using the Product Interoperability Matrix.
- If there any firewalls between the management components review the ports required for SRM in this KB.
- SRM can be licensed in packs of 25 virtual machines, or for unlimited virtual machines on a per CPU basis with vCloud Suite. Read more about SRM licensing here.
- Array based replication or vSphere Replication should be in place before beginning the SRM install. If you are using array based replication contact your storage vendor for best practices guide and the Storage Replication Adapter which is installed on the same server as SRM.
As well as the requirements listed above the following points are best practices which should also be taken into consideration:
- Small environments can host the SRM installation on the same server as vCenter Server, for large environments SRM should be installed on a different system.
- For vCenter Server, Platform Services Controller, Site Recovery Manager servers, and vSphere Replication (if applicable) use FQDN where possible rather than IP addresses.
- Time synchronization should be in place across all management nodes and ESXi hosts.
- It is best practice to have Active Directory and DNS servers already running at the failover site.
In this example we will be installing Site Recovery Manager using Nimble array based replication. There is a vCenter Server with embedded Platform Services Controller already installed at each site. The initial screenshots are from an SRM v6.1.1 install, but I have also validated the process with SRM v6.5.1 and vCenter 6.5 U1.
The virtual machines we want to protect are in datastores replicated by the Nimble array. For more information on the storage array pre-installation steps see the Nimble Storage Integration post referenced below. The Site Recovery Manager install, configuration, and failover guides have no further references to Nimble and are the same for all vendors and replication types.
The installation is pretty straight forward, download the SRM installer and follow the steps below for each site. We’ll install SRM on the Windows server for the primary / protected site first, and repeat the process for the DR / failover site. We can then pair the two sites together and create recovery plans.
Log into the Windows server where SRM will be installed as an administrator, and right click the downloaded VMware-srm-version.exe file. Select Run as aministrator. If you are planning on using an external database then the ODBC data source must be configured, for SQL integrated Windows authentication make sure you log into the Windows server using the account that has database permissions to configure the ODBC data source, and run the SRM installer.
Select the installer language and click Ok.
Click Next to begin the install wizard.
Review the patent information and click Next.
Accept the EULA and click Next.
Confirm you have read the prerequisites located at http://pubs.vmware.com/srm-61/index.jsp by clicking Next.
Select the destination drive and folder, then click Next.
Enter the IP address or FQDN of the Platform Services Controller that will be registered with this SRM instance, in this case the primary site. If possible use the FQDN to make IP address changes easier if required at a later date. Enter valid credentials to connect to the PSC and click Next. If your vCenter Server is using an embedded deployment model then enter your vCenter Server information.
Accept the PSC certificate when prompted. The vCenter Server will be detected from the PSC information provided. Confirm this is correct and click Next. Accept the vCenter certificate when prompted.
Enter the site name that will appear in the Site Recovery Manager interface, and the SRM administrator email address. Enter the IP address or FQDN of the local server, again use the FQDN if possible, and click Next.
In this case as we are using a single protected site and recovery site we will use the Default Site Recovery Manager Plug-in Identifier. For environments with multiple protected sites create a custom identifier. Click Next.
Select Automatically generate a certificate, or upload one of your own if required, and click Next.
Select an embedded or external database server and click Next. If you are using an external database you will need a DSN entry configured in ODBC data sources on the local Windows server referencing the external data source. Click Next.
If you opted for the embedded database you will be prompted to enter a new database name and create new database credentials. Click Next.
Configure the account to run the SRM services, if applicable, and click Next.
Click Install to begin the installation.
Site Recovery Manager is now installed. Repeat the process to install SRM on the Windows server in the DR / recovery site, referencing the local PSC and changing the site names as appropriate. If you are using storage based replication you also need to install the Storage Replication Adapter (SRA) on the same server as Site Recovery Manager. In this example I have installed the Nimble SRA, available from InfoSight downloads, which is just a next and finish installer.
After each site installation of SRM you will see the Site Recovery Manager icon appear in the vSphere web client for the corresponding vCenter Server.
Providing the datastores are replicated, either using vSphere replication or array based replication, we can now move on to pairing the sites and creating recovery plans in Part 3.