With the release of VMware Cloud Foundation (VCF) 9.0, a new architectural construct has been introduced called a VCF Fleet. This brings centralized fleet management and provides modern workload consumption across multiple VCF instances, all managed through a single deployment of VCF Automation and VCF Operations, as shown in the diagram below.
A common question that has been coming up quite a bit lately is how many VCF Instances can a VCF Fleet support?
Technically, there is not a fixed number of VCF Instances that can be managed by VCF Operations. As with with most things in our industry, the answer is "it depends" ...
Before we take a closer look at what "it depends" actually means, it is important to understand what is a VCF Instance?
A VCF Instance includes a VCF Management Domain and can include one or more VCF Workload Domains.
- VCF Management Domain includes the core SDDC components: vCenter Server, ESXi hosts, vSAN or other supported principal storage and NSX Manager) along with single instance of SDDC Manager and a VCF Operations collector.
- SDDC Manager provides lifecycle management for all deployed core SDDC components within a VCF Instance
- VCF Operations Collector provides inventory and metric collection within a VCF Instances, which is then sent to VCF Operations
- VCF Workload Domain, when deployed only includes the core SDDC components and all lifecycle management and operations is performed by the single instance of SDDC Manager
Now that you have an understanding of a VCF Fleet and VCF Instance, lets now look at what attributes can factor into the number of VCF Instances that a VCF Fleet can support?
Unlike traditional configuration maximums which only looks at the number of vCenter Servers, vSphere Clusters and Hosts, VCF Operations uses a different measurement, which is expressed as Objects and Metrics. The deployment size and the number of VCF Operations nodes also plays a critical role in the maximum number of supported objects and metrics.
For example, using the Extra Large deployment size of VCF Operations we can have the following maximums:
- 1 x Node VCF Operations: Up to 100K objects and 20M metrics
- 16 x Node VCF Operations: Up to 1M objects and 126M metrics
Note: Please refer to the VCF Operations section of the VMware Configuration Maximum site for the complete list of object and metric maximums.
If you are running VCF Operations today, you can see the number of objects and metrics that is being collected by navigating to Administration->Control Panel->Cluster Management. In addition, this view only provides cluster management objects and metrics. Each VCF Operations Adapter also provides granular objects and metrics which is a subset within the cluster management view, you will need to navigate to Administration->Cloud Proxies and for a given proxy, you can view the objects and metrics that is collected per adapter.
Network latency and bandwidth also play a critical role, especially with the different communication paths between the VCF Fleet management components and the individual VCF instances.
Hopefully you can now understand why there is not a fixed number of VCF Instances that a single VCF Fleet can support, as there are many factors that can play into the overall calculation and that does not even take into consideration availability or other non-technical requirements that might require you to deploy multiple VCF Fleets.
With all that said, we do want a simpler user experience when it comes to sizing your VCF Fleet and we have a number of active efforts to improve the existing VCF Configuration Maximums site, ports and protocol, sizing recommendations as well as in-product config max visibility.
In the short term, we are also working on a VCF latency document that will cover latency and bandwidth requirements that will aide architects in designing a VCF Fleet, so stay tuned for that resource! I will post the link to the document when it is available.
Thanks for the comment!