Best Practice - Migration VMs with RDMs

[jamiestarr66]

Hello all, new to the channel.
I have a project involving migrating 600 VMs to a new datacenter, 60 VMs has RDM (physical storage) configured. The customer
is leaning toward HCX but is hesitant since this will be a complete move to the new Data center.

Can someone please provide:
1.  What is the best practice for migrating VMs with RDMs?
2.  The customer has a 5gb pipe between the two datacenters to support migration. How do I calculate or determine
     bandwidth utilization?
3.  What would be other options other than HCX to support the migration efforts?
     
     Thanks in advance...

[Christopher Dooks]

Hi Jamie,

1 - Physical RDMs cannot be migrated. Consider converting the Physical RDM into a VMDK or change compatibility mode to virtual. 

2 - Migration performance is down to many factors, such as storage, host resources, bandwidth, packet loss, latency, data churn rate, number of disks, size of disks etc etc. I've seen as much as 1 Gbps on 10 Gbps using vMotion and bulk migration but again, it depends on multiple factors. YMMV, you'd have to perform some testing to be sure.

3 -  Outside of VMware products nothing springs to mind, however do bear in mind products such as SRM and vSphere Replication would have the same Physical RDM limit.

Hopefully this helps.

[jamiestarr66]

Hi Chris..

 

This is perfect, thanks for responding.

[aneeshkem]

Hello  Chris,

We have servers with RDM enabled and are planning to migrate them to AVS. Could you provide the best practices for migrating RDM servers to AVS using HCX tool 4.10

[Magical Grove]

Your project sounds like a big undertaking, but it's manageable with the right approach. Here are some insights to help you out:

1. Best Practices for RDM Migration:

   • For physical mode RDMs, consider converting them to VMDKs before migration if possible, as HCX does not support physical RDMs. For virtual mode RDMs, you can migrate them with HCX, but testing is crucial to ensure application compatibility post-migration.
   • If converting RDMs isn't an option, you'll need to handle storage migrations separately, potentially leveraging storage replication tools or SAN-level replication.

2. Bandwidth Utilization Calculation:

   • Use this formula: (Total data to transfer / bandwidth) = transfer time. For example, if you're migrating 10TB and have a 5Gbps pipe, the theoretical transfer time is about 4.44 hours (assuming 100% efficiency).
   • Don't forget to account for overhead, network contention, and potential throttling. Tools like HCX provide WAN optimization, which can help optimize bandwidth usage. Receiptify

3. Other Migration Options:

   • vMotion: If the latency between datacenters is low, vMotion could be a good choice for live migrations.
   • Cold Migration: For VMs that can tolerate downtime, cold migration with a tool like VMware Converter might be simpler.
   • Replication Tools: Tools like Zerto or Veeam can replicate VMs to the new datacenter, then cut over when ready.
   • Custom Scripts: For RDMs or unique configurations, custom scripts or a hybrid approach may be required.

Lastly, it's critical to document and test your migration process with a small subset of VMs before proceeding with the full migration. Feel free to share more details if you need specific recommendations!