VMware Site Recovery Manager + vSphere Replication Install Guide

Virtualized environments are highly resilient to hardware failures thanks to redundancy and mechanisms such as vSphere HA. VMware Site Recovery Manager solves the challenges related to orchestrating a disaster recovery plan, a major concern for every organization. In this article, we will demonstrate how to set up VMware site recovery manager and get you started on your disaster recovery journey.

A Disaster Recovery Plan, also called DRP, is what may very well save the day should a datacenter wide outage happen. Whether that is flooding, fires, tornadoes, history is plagued with stories of companies going under because they didn’t have a proper disaster recovery plan in place.

Note that when things are fine, Site Recovery Manager is also useful as a planned migration tool to move virtual machines across datacenter with minimal outage. However, this article will focus just on the DR application.

If you want to dive deeper into backup and disaster recovery, download the Backup Bible – at over 200 pages, it’s the complete guide to backup and DR and completely free! Learn more about the Backup Bible

DR Planning, Testing and Documentation

Although this blog is about setting up VMware Site Recovery Manager, I want to quickly touch base on the “soft” topic of the actual organization of a DR. The day things go sideways and management decides it’s time to invoke DR, you don’t want to waste time scrambling through the network shares (which may be down) looking for old documentation, or worse, know that there isn’t one…

These situations are stressful enough and the team’s energy is best focused on following pre-defined steps that you know ensure a successful recovery. An efficient way of achieving this is to follow these steps:

• • Documentation:

• • • Document the recovery steps, communication channels, actors, special cases, etc in a document that is accessible by everyone in the team.

• • • Keep the documentation up to date whenever a change is made. Make it an integral part of your workflows.

• • • Ensure all the members of the team are familiar with the procedure and get VMware site recovery manager training.

• • Testing:

• • • Run planned disaster recovery tests regularly in partnership with the customer (internal or external), at least twice a year to ensure it is working properly.

• • • Get all team members to run DR tests. You don’t want to rely on only one or two colleagues for such a critical piece of work.

• • New VM protection:

• • • Make sure the VM provisioning workflow includes setting up site replication.

Refer to our blog on the subject for more recommendations about disaster recovery, RTO/RPO…

You will also find lots of valuable content in this Q and A session with Andy and Eric.

Site Recovery Manager Components

First off, I want to clarify some confusion that is often brought up between vSphere Replication and VMware Site Recovery Manager. SRM is an orchestrator. It doesn’t perform any sort of replication. The file copy is handled by vSphere Replication (VM) or through storage array volume replication (LUN).

VMware Site Recovery Manager: Download

• • Disaster Recovery orchestrator.

• • Does not replicate virtual machines.

• • Subject to paid license.

• • Can leverage VM-based replication or array-based replication.

VA Type	CPU	RAM	Disk	Network	Capacity
Light	2 vCPU	8GB	20GB	1Gbps	< 1000 VMs
Standard	4 vCPU	12GB	20GB	1Gbps	> 1000 VMs

 

vSphere Replication (VM-based replication): Download

• • Copies virtual machine files to the remote site over the LAN.

• • Relies on vSphere Replication agents on the ESXi hosts.

• • The VR appliance contains the VR management and an embedded replication component.

• • Can be used on its own (license included in most vSphere editions such as Ess+, Standard, Ent+, Desktop…).

Array-Based replication

• • Uses a middle man called Storage Replication Adapter (SRA) which enables SRM servers to communicate with the storage array.

• • Storage Array volumes are replicated using the array’s mechanism over the SAN.

• • Smallest item of granularity is the datastore.

In this blog, we will be using vSphere Replication. And because a picture is worth a thousand words, you can see in the diagram below how the components communicate with each other.

“Component’s interaction in VMware Site Recovery Manager 8.4”

Note that there are several architectures you can put in place in SRM with multiple sites. In this article however, we will stick to the dual “Protected site” and “Recovery site” as this is the most widely spread scenario.

VMware Site Recovery Manager License

Like most VMware products, it comes with an unrestricted 60-days evaluation VMware Site Recovery Manager license. When the license expires you can no longer add VM to existing protection groups or create new ones.

VMware site recovery manager licenses let you protect a specific number of virtual machines and come in a pack of 25 VMs. Note that you need to install a VMware Site Recovery Manager license on each site where VMs are protected.

“VMware Site Recovery Manager license lets you protect a number of VMs.”

The product comes in two editions: Standard for organization that don’t need to protect more than 75 VMs and Enterprise for those that need more or require additional features such as stretched storage support, vvol… See the table below for a comparison of the 2 editions.

“VMware Site recovery manager license comparison”

Network considerations

Bandwidth

Before getting stuck in, you might want to ensure the link between your protected and recovery sites offers sufficient bandwidth for the replication to occur. I would like to recommend the vSphere Replication calculator. A free online VMware tool that will help you estimate your needs in terms of bandwidth, storage, windows etc…

Component communications (TCP ports)

You also want to ensure that the correct TCP ports are open so the infrastructure components can communicate.

Refer to the VMware documentation to get the exhaustive list of network ports:

• • Network Ports for Site Recovery Manager

• • Services, Ports, and External Interfaces That the vSphere Replication Virtual Appliance uses

VM network on recovery site

Although we won’t spend too much time here as this blog is already way too long, this section is probably one of the most important ones. When dealing with workload replication across sites, the network is often the pain point that needs to be addressed quickly and be properly understood.

There usually are 2 options when it comes to it, arguably 3 if you like to live dangerously:

• • Re-IP (IP customization)

If your sites are routed (they should be) and using different subnets, you will need to modify the IP of the VMs as part of the recovery process to put them in the destination network. This usually brings a whole lot of interesting surprises and discussions with the Apps and Dev teams if the services running in the servers don’t fully rely on DNS records (hardcoded IPs) in which case they will surely break.

This model is complicated to maintain in VMware Site recovery manager but works in most legacy network environments.

• • Overlay networks (Network virtualization)

If your organization uses network virtualization techniques such as NSX-T in order to overlay layer 2 networks over routed layer 3, you won’t need to re-IP anything as the same subnet is available in both sites.

This setup makes maintaining site recovery manager very simple at the expense of a more complicated network architecture.

• • Stretched VLANs (don’t do this)

I’m only putting this here because more companies that I would like to admit still do it. You may be tempted to connect an inter-site black fiber to edge switches and span the VLANs across sites? While this sounds like an easy enough solution, it brings issues such as broadcast domains spanning across sites, exposing both datacenter to the consequence of a network loop for instance.

Site Recovery Manager 8.4 Installation and Configuration

Step 1: Deploying Site Recovery Manager Appliance

In this section, we deploy the VMware site recovery manager appliance in both the protected and recovery site.

Note that in our example here, the protected site is called “Site B” and the recovery site is called “Site C”.

• • First, download the appliance from my.vmware.com. Extract the content of the “bin” directory from the ISO file to a folder on your machine.

• • Log into vCenter and Right-click on the inventory object where you want to deploy it then select Deploy OVF Template.

Note that you can deploy the SRM Manager in another vCenter than the one that will be protected (as opposed to vSphere Replication).

• • Check Local file and select the content of the bin directory you extracted from the ISO file and click Next.

• • I will skip some of the steps that are common to all OVF deployments. Here I did set a name and a compute resource beforehand.

• • In the Configuration pane you get to choose the size of the site recovery manager appliance. I obviously selected 2 vCPU for my lab environment.

• • • 2 vCPU: Protect up to 1000 VMs

• • • 4 vCPU: Protect more than 1000 VMs

• • Plan for 20GB of space in the storage section. I also set it as thin but that will depend on your own policy…

• • In the Network section, make sure to select a network that will let you have the right communication fluxes open. Refer to the Network Considerations section for more details on that.

• • You then get to configure the appliance. As usual, make sure the DNS records have been created beforehand and use FQDNs as much as possible.

• • Finally, review the configuration and hit Finish to kick off the deployment of the appliance.

Once the deployment is finished, proceed with deploying the second Site Recover Manager appliance in the recovery site.

 

Step 2: Deploying vSphere Replication Appliance

As mentioned previously, Site Recovery Manager is the orchestrator (the brain), whereas vSphere Replication is the one doing the heavy lifting work of replicating the virtual machines to the recovery site (the muscles).

Note that, as opposed to the VMware site recovery manager appliance, the vSphere Replication appliance must be deployed in the vCenter instance managing the resources to protect/recover.

Unsurprisingly, the deployment process is very similar to the Site Recovery Manager appliance shown above.

• • First, download the appliance from my.vmware.com. Extract the content of the “bin” directory from the ISO file to a folder on your machine.

• • Right-click on the inventory object where you want to deploy it and select Deploy OVF Template.

• • Check Local file and select the content of the bin directory except for the “AddOn” files and click Next.

• • Again, I skipped to the configuration pane where you select 2 vCPU or 4 vCPU. I chose 2 vCPU in my lab but selecting 4 will ensure better performances in a larger production environment.

• • In the Network section, make sure to select a network that will let you have the right communication channels open. Refer to the Network Considerations section for more details on that.

• • Again, go ahead and create the DNS records to use FQDNs for the new server.

• • Leave the vService bindings section as default (not that you can change it anyway).

• • Complete the wizard and hit Finish to kick off the appliance deployment.

Once the deployment is finished, proceed with deploying the second vSphere Replication appliance in the recovery site.

Step 3: Configuring Site Recovery Manager Appliance

The process of configuring the appliance itself is as easy as it gets, as long as the network side of things has been taken care of properly. All we have to do here is to connect the Appliance to its site’s vCenter server.

• • First, open a web browser on the site recovery manager appliance’s configuration page at https://site-b-srm.lab.priv:5480 and log in with admin and the password you set when deploying it.

• • Once you’re in, you can change some of the settings of the appliance should you need to. If everything looks in order, click on CONFIGURE APPLIANCE to connect it to your vCenter.

 

• • In the Configure wizard, you start by connecting the PSC (Platform Service Controller). If you want to use a named service account such as Site Recovery Manager (SRM) with Active Directory, it needs to have SSO admin permissions in vCenter. Here we used the default SSO admin account. When you hit Next you may have to accept the PSC’s certificate’s thumbprint.

Note that we are connected to site-b-srm here so we will be connecting to site-b-vcenter.

• • Next you have to select a vCenter server. If your vCenter servers are connected in Enhanced Linked Mode like mine, select the one that is local to the site you are configuring. You may have to accept the thumbprint of vCenter this time.

• • Finally, enter a name for the site, an email for notifications and leave the rest as default. You would only have to change the other fields in the case of shared recovery sites implementations.

• • Now review the configuration details and hit Finish to complete the connection.

• • The new site and vCenter should now appear in the Summary pane of the appliance and the notification in the top right corner should be successful.

• • Now if you log out of vCenter and log back in, you should see the Site Recovery plugin which wasn’t there before.

Again, go through the same process to configure the appliance on the recovery site (Site C in my case).

Step 4: Configuring vSphere Replication Appliance

Now that the site recovery manager appliance is taken care of, let’s do the same thing with vSphere Replication. I will show the screenshots to offer a uniform write-up but you’ll see that the process is very similar to what we did for the site recovery manager appliance.

• • Open a web browser on the vSphere Replication appliance’s configuration page at https://site-b-vr.lab.priv:5480/ and log in with admin and the password you set when deploying it.

• • Again, you can change some settings if needed, otherwise, head over to CONFIGURE APPLIANCE.

• • Type in the details of the PSC, which is most likely the same as your vCenter since external PSCs are now deprecated, and accept the certificate thumbprint.

• • Then select the vCenter that is local to the site you are configuring. Again, if you don’t have Enhanced Linked Mode, you will only see one vCenter here.

• • Finish by setting up the site name and email address as we did previously and complete the wizard.

• • Just like with site recovery manager, the summary pane should now show the site’s details.

 

When you are done with the first site (Site B here), proceed with configuring the recovery site (Site C).

Step 5: Sites pairing

At this point we have laid down the foundations by deploying and configuring the appliances. Only now will we start configuring site recovery manager itself. The first step of doing so is to pair the protected and recovery sites together.

• • Log in one of the vCenter servers > Home > Site Recovery. If you don’t see this icon, you need to log off and log back in for the plugin to show up.

• • In the Site Recovery pane, you should see both site recovery manager and vSphere replication instances with a green checkmark on all of them. Click on OPEN Site Recovery.

• • This will redirect you to the Site Recovery dashboard where you will manage SRM. Click on NEW SITE PAIR.

• • In the Pair Type window, select the local vCenter Server.

• • • If you use Enhanced Linked Mode like in this screenshot, check “pair … in the same SSO domain”.

• • • Otherwise, leave “pair … in a different SSO domain”.

• • Then you select the remote vCenter server to pair it with.

• • Check both Site Recovery Manager and vSphere Replication checkboxes in the Services pane.

• • If you left the default self-signed certificates, you will be asked for both site recovery manager servers to trust the thumbprint of the remote SRM and vCenter instances.

• • Review the changes by making sure everything is in the right place and hit Finish to enable sites pairing.

The site pair should now appear in the dashboard of Vmware Site Recovery Manager.

Step 6: Resource mappings

We are now at a point where both sites are aware of each other and are awaiting further configuration. We now need to make VMware site recovery manager aware of resource associations from one site to the other, called resource mappings. This will save us from doing it individually for each virtual machine in the protection groups.

We must map folders, compute (named resources), network and storage policies if your organization leverages them.

• • In the main dashboard of VMware site recovery manager, under the new site pair, click on VIEW DETAILS.

This will let you configure everything related to this pair, including mappings.

Network Mappings

VMware Site recovery manager will use the network mappings to know which portgroup to connect recovered virtual machines to in the recovery SDDC.

Note that this process is fairly simple from a site recovery manager perspective. However, it is critical that the portgroup used in the recovery site offers the same gateway and firewall rules as the one in the protected site, unless you use IP customization, which adds a layer of complexity. Refer to VM network on recovery site mentioned earlier for more details on this.

• • Go to Site Pair > Configure > Network Mappings then click on New. Note that Site B (a.k.a. our protected site) is selected by default.

• • In the wizard, you can select to prepare mappings manually or automatically. In Automatic mode, the system will map networks with the same name. A popular choice when many portgroups are involved

 

• • Check the objects you want to map and click ADD MAPPINGS.

• • Check everything in the Reverse mappings section as it will be used during failback.

• • You can select a test network to use when you trigger a test DR. By default, an isolated temporary network will be used. Note that if you do want to use a test network, you obviously need to ensure there will be no IP conflict.

• • Review the mappings configuration and hit Finish.

Folders Mappings

We then move on to mapping the VM folders. The process is very similar to the Network mappings.

• • Go to Site Pair > Configure > Folders Mappings then click on New.

• • Again, you get to choose how to map the resources. This time I chose Manually to show you the difference.

• • Here you will need to map and add them one by one.

• • Apply the reverse mapping here as well and finish the wizard.

Resource Mappings

Last but not least, we will map the compute resources which include resource pools, standalone hosts, vApps, or clusters

• • Go to Site Pair > Configure > Resource Mappings then click on New.

• • This section can’t be done automatically. Select the source and destination resources to pair and click ADD MAPPINGS.

• • Don’t forget to enable the reverse mappings to ensure failback works and finish the wizard.

Step 7: Setting up protection

We have now completed the setup of the basic underlying infrastructure necessary for our Disaster Recover activity. Production environments will obviously have a lot more mappings to go through than this simple lab setup.

This “protection” step includes several things to configure in order for our DR plan to be ready:

• • Configure VM replication.

• • Create Protection Groups (PG).

• • Create Recovery Plans.

This step must be properly thought out from an organizational point of view as this is where you set your RPO and how you actually orchestrate the recovery of your workloads. Before cracking on, we suggest you take some time to define how you want to group virtual machines, what RPO is needed for which VMs and decide an order in the recovery process.

For best practices regarding setting up the protection, refer to VMware’s considerations.

Configure VM replication

Let’s start with enabling the actual replication of the virtual machines.

Note that all the VMs in a replication configuration have the same settings (direction, RPO, schedule…). You will need to create different replications for different configurations.

The available RPO range is from 5 minutes to 24 hours.

• • Still in our “Site Pair” interface, go to Replications, check the direction of the replication and click NEW.

• • The Configure Replication wizard starts with selecting the vSphere Replication Server. Leave the pre-selected Auto-assign unless you need to manually choose a specific one.

• • Next, select the Virtual Machines you want to replicate using this configuration.

• • Select the Datastore at the recovery site and set the appropriate disk format and storage policy.

Select seeds: You can speed up the replication process by physically moving the VM files to a storage medium in cases where the bandwidth between the sites is too small. Once copied, the seeds (files) are detected and avoid replicating a first full.

• • The Replication Settings is very important as we are setting up the Recovery Point Objective (RPO) and general replication such as:

• • • Enable Point in time instances to adjust the snapshots of the source virtual to keep.

• • • Guest OS quiescing for VMs supporting it.

• • • Network compression for VR data to save network bandwidth and help reduce the buffer memory used on the VR server at the expense of CPU usage.

• • • Encryption for VR data. Note that it is automatically on for encrypted VMs.

• • You can specify new or existing protection groups directly from the wizard but we’ll do it manually in the next section.

• • Finally, review your replication settings and hit Finish.

• • You can hit Refresh to follow the VM copy. At the end, the replication should be successful.

• • If you look inside the datastore at the recovery site, you will find a folder with the replicated VM files in it.

Create Protection Groups

A protection group (pg) is a collection of virtual machines that VMware Site Recovery Manager protects together. There are different types of pg with array-based replication, VVOLs, vSphere replication, however, they can be mixed together in a recovery plan (more on that later).

With protection groups, VMware Site Recovery Manager will create placeholder VMs in the recovery site with the right mappings, which is part of the upside of using Site Recovery Manager and not just vSphere Replication on its own.

Disclaimer regarding protection group design: In this example, I created a pg for the web nodes and one for the database in order to show you 2 different groups in the UI. However, in real life you would more likely either create a pg per application stack for high granularity or wider ones, at the department level, for instance, to reduce complexity.

• • Browse to Protection and click NEW.

• • Give your protection group a name and a description that quickly and clearly identify the type of workloads it protects. You also need to choose the direction of the protection.

• • In the Type section, select Individual VMs (vSphere Replication).

• • Select the virtual machines to add to the protection group.

• • In the Recovery Plan section, I created a new plan named SaaS stack.

• • The protection groups should now show up in the UI.

Again, creating different protection groups for web and db nodes don’t usually make much organizational sense so you want to really think this through with your peers.

Create Recovery plans

In VMware Site Recovery Manager, recovery plans are flexible, customizable and automated run books that orchestrates the recovery process (VM start/stop order, network addresses to use…).

• • A recovery plan includes one or more protection groups.

• • A protection group can be part of multiple recovery plans.

In the screenshots below we already created a recovery plan in the protection group wizard so we’ll see how to configure it here.

• • Go to Recovery Plans > “RP Name”. You can see in the Recovery Steps the order of the recovery scenario. In our case, we want to start the DB before the web nodes.

• • You should find your protection groups in the Protection Group tab. If not, add them.

• • Now click on Virtual Machines > Select the VMs to start first (databases in our case) > right click > Priority Group > 1 (Highest).

• • You can then configure the start-up order you want for the other VMs in the list.

• • You can go further if you click on Configure Recovery in the context menu, you will get additional options for the VM such as VM dependencies, power options as well as IP customization (a.k.a. re-IP).

More information about IP customization in the VMware Site Recovery Manager documentation.

 

Step 8: Testing Recovery Plan

Now that the Recovery Plan is created, it is time to test it. As we mentioned earlier, it is important to test your disaster recovery infrastructure regularly to ensure it works when you need it.

Keep in mind though, that the Test functionality of the site recovery manager will only start the VM and check for the VMware Tools by default. This is fine to check that SRM works, but it is not enough in a production environment. You should schedule a dedicated maintenance window to recover (not test recover) the production workloads at least twice a year.

Testing recovery plans does not impact production as long as the test network doesn’t incur an IP conflict.

Note that the network used for the test is the one specified in the network mappings section, which by default is isolated.

• • In the Recovery Plans tab, make sure the plan is ready for test or recovery and click TEST.

• • In the wizard you can choose to replicate the changes that occurred since the last replication. Checking it simulates a planned failover while leaving it unchecked simulates a disaster recovery scenario where the source VM is unavailable.

• • There isn’t much to review in the second window, click on Finish to start the Test recovery plan.

• • You can monitor the progress of the job in the recovery plan page and follow what action is being performed. Note that, by default, it will wait for the VMware Tools before proceeding to the next step.

If you want to skip the VMware Tools check you will need to configure it on the VM in the recovery properties.

• • Once the job is finished, you can look in the vCenter at the recovery site and you will find the recovered virtual machines in running state. Again, the source VMs are not impacted and remain running.

• • Once you are happy with your DR test, hit the CLEANUP button to remove the test environment and reset the plan to ready state.

• • Click Next in the Cleanup wizard.

• • Click Finish to run the cleanup operation and complete the test scenario.

Step 9: Executing Recovery Plan

In the previous test section, we started a copy of the protected VM in an isolated network in the recovery environment while the source VM was kept running. Now we will execute the recovery plan which will switch the workload to the recovery environment.

Note that there are 2 methods for Site recovery manager to execute a recovery plan:

• • Planned Migration: Replicate recent changes to the recovery site and cancel recovery if errors are encountered.

• • Disaster Recovery: Attempt to replicate recent changes to the recovery site, but otherwise use the most recent data. Continue recovery even if errors are encountered.

In this example, we will be performing the planned migration.

• • On the recovery plan page, make sure the status is Ready and click on the RUN button.

• • Here you need to check a box to ensure that you understand what you are about to do, i.e., shut down the production workload and switch over to the recovery environment. Here we select the Planned Migration scenario and click NEXT.

• • Click Finish to launch the operation. I shall reiterate that there will be an interruption of the service for the duration of the switcheroo.

• • Once the operation is completed, you will notice in vCenter that the source virtual machines are switched off.

• • We ran a ping for the duration of the recovery which took less than 2 minutes for 2 Ubuntu VMs in order to visualize the actions performed automatically. Now keep in mind that this is the most basic of use cases so don’t expect your DR to run in 2 minutes.

• • You should also find the plan status on Recovery complete. Notice the warning about your workloads not being protected. This is because there is no protection in the other direction yet.

Step 10: Re-Protect and failback

When a recovery plan is executed, the workloads are switched to the recovery site but what happens when you want to go back to normal and move them back to their original site?

This is known as the failback process which consists of making the recovered VM into the protected virtual machine (the protected site becomes the recovery site) and execute the recovery plan in the other direction.

Reprotect

• • In the recovery plan window, click on the REPROTECT button in the warning banner.

• • Again, you need to check the box to make sure you understand this operation cannot be undone.

• • You can then click Finish to launch the Reprotect operation.

Note that the Reprotect operation has no impact on the state of the VM. It simply turns the original protected VM into the replica. You will actually notice that the icon in vCenter changes to the one with 3 squares next to it.

Failback

The process of failing the virtual machine back to its original site is exactly the same as executing a recovery plan because it is just that.

Exporting reports

VMware Site Recovery Manager includes an export functionality to extract the operations in various formats such as html, xml, csv, xls and doc. That way you can provide your management or clients a proof that a test or actual recovery has been performed.

• • Go to Recovery plan > History > Export all.

• • Select the format that works for you, we chose html below and click Download or Open in new tab.

• • The looks of the report is a bit simplistic but it includes the core information you are most likely after.

Is this Sufficient?

Although this was a pretty lengthy article, we have only scratched the surface of what VMware Site Recovery Manager can do to orchestrate your disaster recovery or planned migrations, especially in the recovery plan options and parameters. You can also get VMware Site Recovery Manager training on the VMware hands-on labs.

Whilst VMware Site Recovery Manager will be suitable for basic and non-essential workloads for more important environments such as business dependent applications you will need something more comprehensive and reliable. That’s why most companies turn to third-party vendors who specialise in these types of software. If you feel like VMware site recovery manager doesn’t quite fit the bill for you in terms of technical features or price point, make sure to check out Altaro VM Backup and Replication.

You may also want to have a look at setting up some automation to reduce the operational overhead of maintaining VMware Site Recovering Manager by leveraging the SRM PowerCLI module.

In this article, we only covered the on-premise aspect of VMware Site Recovery Manager but be aware that VMware offers cloud-based disaster recovery solutions such as VMware Site Recovery and DRaaS.

Again, if you are new to the disaster recovery terminology, we strongly suggest you make sure you understand RPO/RTO and draw up a disaster recovery plan before starting configuring your replication, protection groups, recovery plans etc… For everything you need to know about backup and DR, grab your free 200+ page copy of The Backup Bible.