When do I create a WLD and when do I create a cluster?
This is a subject that entirely depends on the strategy or the requirements of the customer, but I will outline why this is important and what needs to be considered.
- Scale –
- There is a limit on the number of workload domains
- Each workload domain requirements an additional vCenter. The vCenter will need to be managed.
- Each workload domain requires its own lifecycle action/event
- Large workload domains can take a long time to lifecycle (more recent VCF releases assist in this)
- Fault domain design:
- Does building a WLD provide the required number of fault domains/failure points that is accepted by the SLA/KPI/Requirement for availability.
- Can a cluster provide the required number of fault domains/failure points that is accepted by the SLA/KPI/Requirement for availability?
- Workload licensing may require a separate WLD with its own management point (vCenter).
- Licensing for VCF?
- First of all what does “multitenancy” mean to the organization? Every organization appears to have a separate definition.
- Physical separation or compute, network and storage would drive a separate WLD
- Separation of storage only could result in a separate cluster?
- Separate NSX control could result in a separate cluster (unless NSX-T is deployed in a consolidated fashion)
These are some of the common considerations when choosing if a WLD or a cluster is the correct decision.
Why is this important?
As detailed above each option has a number of considerations. Choosing the correct option will allow for VCF to be used effectively. If the default is create a new WLD, as this is how the organization has always done this, then the limits of VCF will be hit quickly and the VCF may become more difficult to manage.
There is also a possibility that if the requirements are not understood and the correct option is not selected, that the end user may have a bad experience due to the selected option not meeting their requirements.