Not a zero-downtime option: with this VCF Operations HCX method, a live VM is copied through host-based replication and then, at an optionally scheduled switchover, the original is shut down, final changes are synchronised and the replica started, costing roughly a reboot's worth of outage.
Also called HCX Bulk Migration.
Read more: Broadcom TechDocs
In the Ultra Transcenders books
Each book explains Bulk migration in context, with comparison tables and the common traps.
Terms in this definition
- VCF Operations HCX
Add-on licensed for moving workloads inside and between VCF private clouds and linking those environments, offering bulk or live migration plus network extension so IP addresses can stay the same.
- Azure AI Search replica
Additional copy of a search service's engine and indexes. Replicas spread queries for load balancing and improve availability, so adding them is the fix when queries are throttled with HTTP 503.
Related terms
- guest OS customization
When cloning or deploying from a template, giving the new VM's OS its own name, network and licence details so it does not collide with others. VCF Operations HCX can apply it in Bulk Migration too.
- Mobility Optimized Networking
Avoids tromboning on networks extended by HCX: migrated VMs route via the destination site instead of back through the gateway at the source. This Network Extension option applies automatically to VMs moved by Bulk migration, whereas vMotion and RAV leave VMs on the source gateway until someone reconfigures them.
- NKP
vSphere Native Key Provider is part of vCenter and issues the keys for VM Encryption and vTPM, so no outside key server (KMS) is required. Moving encrypted VMs by HCX bulk migration needs the target to offer a corresponding KMS or NKP.
- Supervisor onboarding
Migration type in HCX for moving traditional VMs into a Supervisor namespace, via Bulk migration to staging followed by an import. It requires HCX 9.0 on both sides and VMs at hardware version 17+ with VMware Tools; as each VM restarts, downtime-intolerant workloads should avoid it.