This section describes how to deploy, configure, and undeploy PrimaryIO Hybrid Cloud Data Management (HDM) 2.2 for enabling migrations of virtual machines to IBM Cloud for VMware Solutions Shared (VCD).
HDM Deployment Steps
HDM deployment involves the following six steps, each of which are discussed in detail in the sections below:
- Deploy the HDM Appliance On-Premises
- Add an On-Premises vCenter to the Appliance
- Register the HDM plugin to the On-Premises vCenter
- Add the HDM License
- Deploy HDM On-Premises
- Deploy HDM in the Cloud
Step 1: Deploy the HDM Appliance On-Premises
The HDM Appliance manages on-premises and cloud HDM operations. This section covers the following topics for deployment and configuration of the HDM Appliance:
- Placement of the HDM Appliance on-Premises
- System Requirements for the Appliance
- Network Requirements
- Deployment and Configuration
- Network Configuration for the Appliance
- Adding vCenter
Placement of HDM Appliance
- The HDM Appliance must be deployed on the same on-premises vCenter it will manage.
- It should ideally be kept in a separate cluster than the one it will manage.
Deployment and Configuration
The HDM appliance must be installed from the OVF in the HDM package. Be sure to check that the placement and sizing of the deployment conforms to the requirements mentioned above. To deploy the appliance, use the “Deploy OVF Template” option in vCenter (figure 6).
Figure 6: Launching the “Deploy OVF” Template in vCenter
Appliance network configuration requires that the available networks be categorized into four types, described in the Network Requirements and HDM Resource Requirements sections, above. While it is recommended to keep these networks separate, it is possible for each of them to point to a single network. In addition, the decision must be made at this point to deploy using a static IP or DHCP protocol (figure 7).
Figure 7: Configuring the Network
IP Address Details
(If DHCP is chosen, this section can be skipped)
For each network type, choose the IP allocation protocol: static or DHCP. The following details are required if static IP is chosen (figure 8):
- IP range: list of IP addresses that are free to be allocated to various HDM components
- Subnet mask: subnet of the IP range
- Gateway: gateway IP of the network
- Domain: domain name of the network
- DNS: domain name translation for the IP address
- NTP: server that will be used for time synchronization
- If static IP is chosen, failure to specify an NTP server may cause further HDM operations to fail.
- If static IP is chosen, the gateway IP and the specified IP range must be in the same subnet.
Figure 8: Configuring the System for Static IP Protocol
The setting in Default gateway network should be chosen so
that traffic for other subnets can be routed through this interface. This would
generally be the WAN network but please verify the specifics for
In the field Default Gateway set the default gateway for the
network chosen under Default gateway network.
Getting this wrong will cause deployment to fail eventually and you will have to perform an HDM Reset
Figure 8b: Configuring Gateway, NTP and Default Gateway
Power on the Appliance
Complete the rest of the wizard and wait for the deployment to complete. Once the deployment has completed, use vCenter to power on the appliance.
Please note that the appliance can take between 5 to 10 minutes to boot depending on the underlying datastore's performance.
Enable VMware HA on the HDM Appliance
(This step should only be performed for Cluster mode deployments of HDM)
VMware HA should be enabled on the cluster where the HDM appliance has been deployed. Follow these procedures to do this:
- In the on-premises vCenter, select the cluster where the HDM appliance has been deployed
- On the right hand pane, select the Configure tab, then select Services followed by vSphere Availability
- If the vSphere HA is turned off, select EDIT and enable vSphere HA
Step 2 : Add On-Premises vCenter to the HDM Appliance
Note: The on-premises vCenter should be added to the same location where the HDM appliance has been deployed.
- Ensure that the HDM appliance is powered on and you have valid vCenter administrator credentials. To do this, log into the PrimaryIO appliance by accessing https://<appliance ip> in a supported web browser and enter the administrator credentials. The default administrator username and password are ‘administrator’ and ‘2Hdm$aK!N@h!’, respectively. This default password should be changed immediatly following the first login.
- Select vCenters from the top pane.
- Select Add vCenter from the top right (figure 9).
Figure 9: Adding vCenter to the HDM Appliance
- Specify the IP or FQDN where the on-premises vCenter is configured. Enter the administrator credentials and select Add vCenter.
- The on-premises vCenter should now be listed on the PrimaryIO appliance
Figure 10: Viewing the On-Premises vCenter on the HDM Appliance
Step 3: Register the HDM Plugin to vCenter
- Select Register to register the HDM plugin to the on-premises vCenter.
- Logout and login twice (occasionally, more than twice will be required) from vCenter for the plug-in installation to complete. Successful installation will add an HDM icon to the vCenter home page.
Figure 11: HDM Icon on the vCenter Home Page
Note: The PrimaryIO vCenter plugin will be automatically registered on all vCenter instances that are configured using enhanced linked mode.
Step 4: Add HDM License
A license key is required to activate the newly-deployed HDM. A trial license will be shared with the product which can be used to activate HDM.
Note: The HDM configuration page may take a few minutes to load, as product settings need to be initialized.
Steps to add a license:
- In the on-premises vCenter, access the HDM dashboard by selecting HDM from Shortcuts or the vCenter menu (It may take a few minutes for the dashboard to load).
- Select the _Administration_tab.
- Select Licensing to add a new license, and to see existing licenses (figure 12).
- Select Add License to open a popup where the provided Trial License can be pasted (figure 13).
- Copy and paste the shared license key and select OK. The license will now be listed on the license page (figure 14).
- The same interface can then be used to add the actual Product License obtained as part of the ordering process.
Figure 12: Add or See HDM Licenses
Figure 13: "Add License" Pop-up
Figure 14: Verify the Addition of the Trial License
Step 5: Deploy HDM On-Premises
This section covers the steps for completing the deployment of HDM on-premises. This section will explain how to:
- Select a cluster for HDM deployment
- Select HDM deployment mode
- Select HDM resource allocation
- Configure the network
- Complete the deployment
The on-premises deployment may take anywhere between 10 and 20 minutes, depending on the network bandwidth and storage. Any failure-retries performed by HDM during the deployment process are logged into vCenter events. Either the hdm or the com.primaryio keyword can be used in vCenter Event Search to search for events generated by HDM.
It is assumed that the HDM appliance deployment and configuration has been executed successfully. Ensure that the plugin has been registered on vCenter, the HDM icon is visible on the vCenter home screen, and a valid HDM license has been added to activate the product.
Steps for deployment:
- In the on-premises vCenter, select PrimaryIO from Shortcuts or from Menu to access the PrimaryIO dashboard
- Select Administration from the tab
- Select Configuration to access the on-premises cluster listing
Figure 15: vCenter On-Premises Cluster Listing
- Select the cluster where HDM will be installed and select Install. This will open a wizard to select resources for HDM component deployment.
- Verify that the Appliance IP is correct.
- Select the deployment mode based on the scale and failure tolerance requirements for the on-premises and cloud environments, taking care to ensure there are enough resources available in the on-premises and cloud environments. See the Deployment Requirements section for details on each deployment mode.
- Select resource allocation based on size and resource requirements. See the Deployment Requirements section for details on each allocation type.
Figure 16: HDM Installation Wizard
- Select the resource pool and shared datastore where the HDM components will be deployed.
Figure 17: Resource Pool and Shared Datastore Selection
To specify the internal network configurations, start by choosing the IP allocation protocol: static (figure 18) or DHCP. The following details are required if static IP is chosen:
- IP range
- Subnet mask
See the Deployment Requirements section for additional details on these requirements. These steps must be repeated for the following components:
- HDM Internal Network (HDM_Internal_Network)
- WAN Network (Uplink_WAN_Network)
- ESXi Network (ESXi_Network)
- If all are mapping to a single network, separate IP ranges will still need to be provided for each network type.
- The subnet range 172.17.0.0/16 is not available for the deployment; it is internally used by HDM microservices.
Figure 18: Network Configuration Settings
- Confirm the selection and submit the installation request by selecting FINISH.
Figure 19: HDM Installation Summary
- This will create a deployment task which can be monitored from vCenter tasks. Also, the deployment progress can be seen in the HDM configuration view:
Figure 20: vCenter HDM Configuration View
Step 6: Deploy HDM to the Cloud
Depending on the WAN bandwidth and WAN latency, it may take between 30 to 60 minutes to deploy HDM components in the cloud. Any failure-retries during the deployment process will be logged into vCenter Events. To search for events generated by HDM, hdm or com.primaryio keywords can be used in vCenter event search.
- Ensure that the HDM plug-in has been registered on vCenter and the HDM icon is visible on the home screen.
Ensure that the on-premises deployment has completed successfully.
- HDM_INTERNAL_NETWORK and WAN Network should be connected to the selected vApp.
Steps for deployment
- In the on-premises vCenter, select HDM from Shortcuts or Menu to access the HDM dashboard
- Select Administration from the tab
- Select Clouds to access the cloud listing
Figure 21: vCenter Cloud Listing
- Select Add Cloud. This will open a wizard where you can select the resources needed to deploy the HDM components.
Select the cloud type, then enter the cloud credentials and the cloud DNS.
Refer the details gathered during deployment planning, as mentioned in "Deployment planning > Prepare Cloud for Deployment" section.
Figure 22: HDM Cloud Wizard
Select vDC, vApp where HDM components will be deployed on cloud and then proceed.
- Select the network and IP configuration for ON_CLOUD_HDM_NETWORK (HDM Internal Network) that will be used by HDM for VM boot and subsequent IO handling through the HDM components. Choose the IP allocation protocol: static (figure 24) or DHCP . If static IP address allocation is chosen, the unused IP addresses or IP ranges, along with the gateway and subnet mask, would need to be specified.
Figure 24: Network Configuration
- Select network IP configurations for the WAN. Choose the IP allocation protocol: static or DHCP (figure 25). If static IP address allocation is chosen, the unused IP addresses or IP ranges, along with the gateway and subnet mask, domain search path, and one or more DNS and NTP servers, would need to be specified
Figure 25: WAN Configuration
- Set WAN route configuration, so data traffic between the cloud and on-premises environments is possible. See the network planning section WAN Routing Configuration.
Figure 26: WAN Route Configuration
Figure 27: Network Mapping
- Select network mapping to enable the network categorization performed on-premises gets percolated to the cloud
- To submit the request and initiate the cloud deployment, select FINISH. This will create deployment tasks for the cloud components, which can be monitored in vCenter Tasks. The progress can also be seen in the HDM Clouds view (figure 29)
After you submit the request, the cloud deployment can take from 45 minutes to
over an hour, depending on storage and network performance. Please note
that the progress bar displayed during this process is not linear and
specific steps take longer, resulting in the progress bar spending
longer time at a certain percentage.
For the Cloud deployment task, it takes about 40-50 minutes at 1%.
Under the Organization VDC upload ISO task, stays at 50% for
about 25 minutes.
Under the Organization VDC upload OVF task, lingers at 1% for about 20 minutes.
Figure 28: Summary and Request Submission
- This will create deployment tasks for the cloud components. These tasks can be monitored in vCenter Tasks. The progress can also be seen in the HDM configuration view
Figure 29: HDM Clouds View
Once the on-premises and cloud deployments are complete, there are multiple ways to view the deployed components and their states:
After successfully installing HDM in the cloud, the appliance control panel will list each component and their health.
Figure 30: HDM Appliance Control Panel
In addition to the Appliance Control Panel, this information is also available on vCenter on the HDM plugin by selecting Menu, followed by HDM, Administration, HDM Health, and then Component Health.
Figure 31: vCenter HDM Plug-In
To verify that the on-premises vCenter shows the cloud state as installed, select HDM, then the Administration and _Clouds _tabs.
Figure 32: Verify Installation of the Cloud State
NOTE: Configuration information needs to be gathered and loaded from vCD, so it may take a few minutes to load the Cloud Dashboard for the first time.
Enable IO Monitoring
- This step is not required for cold migration.
- For Lite Standalone and Lite Cluster deployment modes, this step only creates the SPBM policy and attaches to VMs in the cluster; it doesn’t perform any detailed I/O profiling.
HDM requires the on-premises VMs to be monitored for I/O by HDM. Monitoring helps to:
- Estimate the cloud cache size requirement for a given VM
- Determine whether the VM is a suitable candidate for migration
This is done by creating an HDM SPBM cache type policy and applying that policy to the existing VMs. Detailed steps are provided below.
- Ensure the on-premises installation has successfully completed.
- Ensure there are no HDM SPBM policies named ‘HDM Analyzer Profile’ already existing from previous HDM installation attempts. This can be checked using the following steps:
- In the on-premises vCenter, select Home, followed by Policies and Profiles, then VM Storage Policies
- Delete any instances of ‘HDM Analyzer Profile’
- In the on-premises vCenter, select HDM from Shortcuts or from Menu to access the PrimaryIO dashboard
- Select Administration from the tab.
- Select Configuration to access the cluster listing.
- On the cluster where HDM is installed, select Enable under monitoring. This will open a popup.
- Select OK to enable monitoring.
This will create an SPBM cache type policy for HDM. The default name for the policy is HDM Analyzer Profile. This profile can be viewed in the on-premises vCenter by selecting Home, followed by Policies and Profiles, then VM Storage Policies.
This newly created HDM policy will be applied to all virtual machines under the cluster, and will enable monitoring for I/O activity. A reconfiguration task will be generated for each virtual machine when applying the HDM SPBM policy. If VMs have been successfully reconfigured, VM policies for these virtual machine will have ‘HDM Analyzer Profile’ applied against each of their VMDKs.
Once the policy has been applied, the user can check analyzer summaries by selecting VM->Monitoring->PrimaryIO from Host and Clusters view.
1.The Enable IO Monitoring step can be executed any number of times to enable monitoring for new VMs since the last execution.
- Applying the SPBM policy can fail for some VMs. For example, when the VM already has a snapshot. However, this doesn’t fail the entire operation.
Disabling Monitoring on a VM
Note: This step must not be executed for a VM that will be migrated to the cloud.
Monitoring a VM is a low overhead activity and is done transparently without affecting the ongoing I/Os or operations. However, if you don’t want HDM to monitor certain VMs, take the following steps:
- In the on-premises vCenter, right click on the VM you do not want to be monitored.
- Select VM Policies, followed by Edit VM Storage Policies.
- In the popup, if the VM storage policy is the HDM Analyzer Profile, set the VM storage policy to Datastore Default, then select Apply to all
- The vCenter task will display the operation progress and status.
Note: HDM cannot migrate a VM that does not have an HDM SPBM policy. However, the SPBM policy can be applied to a VM at any time, thereby making it available for migration.
Troubleshooting Common Failure Scenarios
Common failure scenarios and possible reasons are provided below. If the failure is not one of the above reasons and the error message is not clear enough, please contact PrimaryIO Support.
Here are some common reasons for deployment failures:
- Resources not sufficient: On-premises or cloud resources are insufficient for meeting the deployment mode requirements.
- Unresolved input values: Inputs such as vCenter FQDN/IP are not resolvable or are unreachable.
- IP range issues: For static IP allocation, the IP range specified is insufficient for assigning to all HDM components.
- HDM component failed during or after the deployment: HDM component failed during deployment.
Usually when the deployment UI wizard or vCenter tasks fail, one of the above error messages will appear. After the error has been rectified, a re-deployment can be initiated, or an HDM reset followed by re-deployment can be attempted.
Common reasons for undeployment failures are:
- VMs are still migrated to the cloud when undeployment is attempted.
- The on-premises cluster host is not able to go to maintenance mode. This can be due to:
- DRS not enabled on the cluster, causing powered-on VMs to block the maintenance mode
- In the case where the HDM applicance is part of a single host in the cluster, the maintenance mode will be blocked if the HDM appliance is powered on. In this case, uninstall the HDM filter and immediately power off the appliance.
vCenter tasks will provide the reasons for undeployment failures. If it is one of the above, rectify the problem and retry the operation.
Support and Contact
HDM support can be reached via email: email@example.com
The email should include the following information:
- Email subject: Clearly specify the issue.
- Email body: Clearly describe the problem. Include information on the deployment type, the operation being performed, any relevant screenshots, etc.
Email attachment : Attach the relevant logs (as described in HDM Support Logs section)
If the attachment size is too large, an alternate shared location may be provided by HDM support to enable the logs to be uploaded.