Managing Multiple vCenters: The Manual Operation Nightmare
After over six months of operating a VMware cluster with 8 hosts split between two sites in Hanoi and Ho Chi Minh City, I was fed up with constantly having two browser tabs open. Every time I needed to check resources, I had to log in repeatedly. It was incredibly frustrating when I needed to perform a vMotion of virtual machines between sites or quickly assign permissions to new employees.
That is when vCenter Enhanced Linked Mode (ELM) became my lifesaver. Simply put, ELM connects vCenter Servers into a single Single Sign-On (SSO) domain. You only need to log in once to see your entire resource inventory across all sites on a single interface.
Three Approaches for Multi-Site Infrastructure: Which is the Best Choice?
Before configuring, I weighed three options to see which was the most practical solution:
- Standalone vCenter: Each site is an isolated island. This is easy to set up but extremely difficult to manage as the system scales.
- Enhanced Linked Mode (ELM): vCenters share a common SSO “residency” (users, permissions, licenses). This is the gold standard for businesses with two or more sites.
- Hybrid Cloud Link: Connecting on-premise vCenter with the Cloud (like AWS). This solution is too expensive and complex if you are just running a Lab or operating a medium-sized enterprise.
Why I Chose ELM for an 8-Host System
After the transition, administrative efficiency improved significantly thanks to these practical advantages:
- True Centralized Management: Searching for VMs, checking logs, or viewing performance charts are all contained within one screen. No more getting confused between tabs.
- Cross vCenter vMotion: I can drag and drop VMs or copy ISO files between Datastores at two different sites seamlessly.
- Consistent Permissions: I only need to create a User Group on the SSO, and those permissions automatically sync across all vCenters in the group.
A small note: ELM is not a High Availability (HA) solution. If the SSO domain goes down, you will have trouble logging into all vCenters. Therefore, never forget to back up your vCenter regularly.
Mandatory Checklist Before Deployment
I once had to reinstall everything from scratch just because I was careless about the Network. To prevent ELM from failing midway, you need to prepare:
- DNS Must Be Correct: Each vCenter needs accurate A and PTR records. DNS errors are the cause of 90% of domain join failures.
- NTP (Time): The time skew between nodes must not exceed 30 seconds. It is best to point all nodes to an internal NTP Server to ensure consistency.
- Versions: Keep your vCenters on the same version (e.g., both on 8.0 Update 2) to avoid annoying compatibility errors.
- Connectivity: Latency between sites should be under 10ms. Minimum bandwidth should reach 100 Mbps for stable SSO data synchronization.
Steps to Set Up vCenter Enhanced Linked Mode
Scenario: I have vCenter-HN (Hanoi) currently running. Now I will install an additional vCenter-HCM (Ho Chi Minh City) and connect it to the existing system.
Step 1: Configure the First vCenter (Primary)
Install vCenter at the Hanoi site as usual. In the SSO section, select “Create a new vCenter SSO domain”. The default domain name used is typically vsphere.local.
Step 2: Connect the Second vCenter to the System
When installing the vCenter in Ho Chi Minh City, proceed to Stage 2. This is the critical step:
- Select: Join an existing vCenter SSO domain.
- Partner FQDN: Enter the address of the Hanoi vCenter (e.g.,
vcenter-hn.company.com). - SSO Domain name: Enter
vsphere.local. - Username/Password: Use the
[email protected]credentials from the Hanoi site.
The system will take about 10-15 minutes to perform the “handshake” and sync data. Afterward, you just need to refresh your browser.
Tips for Checking and Troubleshooting Sync Errors
Sometimes the interface shows green, but the underlying data is out of sync. To be sure, I usually SSH into the vCenter Appliance and use the following commands:
View the list of linked nodes:
/usr/lib/vmware-vmdir/bin/vdcrepadmin -f showservers -h localhost -u administratorCheck replication status (this command is crucial for detecting out-of-sync nodes):
/usr/lib/vmware-vmdir/bin/vdcrepadmin -f showpartnerstatus -h localhost -u administratorIf a vCenter is severely corrupted and beyond repair, use the following command to cleanly remove its information from the system:
cmsso-util unregister --node-pnid vcenter-hcm.company.com --username [email protected]Hard-Earned Lessons After 6 Months of Operation
Managing an 8-host cluster with ELM has made my life much easier, but there are a few points to keep in mind:
- Don’t Overdo the Quantity: VMware allows linking up to 15 vCenters. However, if the inter-provincial connection is unstable, the vSphere Client interface will load very slowly.
- Snapshot Before Changes: Before removing a node or modifying SSO configuration, take a Snapshot of all vCenters. Manually fixing SSO domain errors is a true nightmare.
- Organize Folders by Site: Even though everything is on one screen, I still organize VMs into separate Folders (HN_VMs, HCM_VMs). This helps avoid accidentally shutting down a colleague’s VM at another site.
Deploying ELM is not difficult. The challenge lies in maintaining the stability of DNS and the underlying network infrastructure. Once everything is running smoothly, you will find managing a virtualized infrastructure much more professional and manageable.
