System Recommendations and Requirements

For optimum performance, you must match the minimum recommendations for the deployment.

Recommendations for the Platform Deployment

Table 1. Specifications for Platform Brick Size
Brick Size	Cores required for 2.1 GHz CPU	Cores required for 2.3 GHz CPU	Cores required for 2.6 GHz CPU	RAM	Disk
Medium	10	9	8	32 GB	1 TB
Large	15	14	12	48 GB	1 TB
Extra Large	20	18	16	64 GB	2 TB

Note:

The reservation for the CPU speed and RAM for each node must be 100% of the value specified above.
To match your setup to all the specifications, you might have to add the resources (RAM, Disk, CPU). See https://kb.vmware.com/s/article/53550 and Increase the Brick Size of your Setup.

Table 2. Non-Cluster Deployment - Maximum Capacity
Brick Size	*Number of VMs (in thousands)	Flows per Day (in millions)	Total Flows (in millions)	Flow Planning (in millions)	Number of Devices	Number of Routes (in thousands)	*Number of Edges for VMware SD-WAN (in thousands)
Medium	4K	1M	4M	2M	10	2K	2K
Large	6K	2M	8M	4M	20	4K	2K
Extra Large	10K	2M	8M	4M	30	5K	4K

Note:

The Network Verification and Assurance feature (Network Map and Intents) is available only for Extra Large brick size only.
* The count of VMs and edges mentioned in the table is the maximum individual limit for a single deployment. So, if you have edges in your setup, you might have to reduce the VM count.
The count of VMs includes the templates on the vCenter as well.
Total Flows is the maximum count of flows the system can store for the retention period.
Flow Planning is the total flows for which the system can perform security planning.

Table 3. Cluster Deployment - Maximum Capacity
Configuration	Brick Size	Cluster Size	*Number of VMs (in thousands)	Flows per Day (in millions)	Total Flows (in millions)	Flow Planning (in millions)	Number of Devices	Number of Routes (in thousands)	*Number of Edges for VMware SD-WAN (in thousands)
Up to 50 Devices	Large	3	10K	2M	8M	4M	30	5K	4K
	Extra Large	3	18K	6M	24M	4M	50	10K	6K
	Extra Large	5	30K	10M	40M	4M	50	10K	10K
	Extra Large	10	100K	15M	55M	4M	50	10K	10K
Greater than 50 Devices	Extra Large	3	12K	3M	12M	4M	300	150K	6K
	Extra Large	5	18K	6M	24M	4M	400	275K	10K
	Extra Large	10	72K	13M	50M	4M	400	275K	10K

Note:

The Network Verification and Assurance feature (Network Map and Intents) is available only for Extra Large brick size only.
The number of routes includes all forwarding entries, including layer 3, layer 2, access control and NAT.
* The count of VMs and edges mentioned in the table is the maximum individual limit for a single deployment. So, if you have edges in your setup, you might have to reduce the VM count.
The number of VMs includes the templates on the vCenter as well.
Cluster size is the total number of nodes in the cluster.
Total Flows is the count of flows in the system for the retention period.
The query to determine the Total Flows is count of flows in last 31 days, assuming the retention period as 31 days.
Flow Planning is the total flows for which the system can perform security planning.

Table 4. Specifications for Collector Brick Size
Brick Size	Cores required for 2.1 GHz CPU	Cores required for 2.3 GHz CPU	Cores required for 2.6 GHz CPU	RAM	Disk
Medium	5	5	4	12 GB	200 GB
Large	10	9	8	16 GB	200 GB
Extra Large	10	9	8	24 GB	200 GB

Note: The reservation for the CPU speed and RAM for each node must be 100% of the value specified above.

Table 5. Collector Deployment - Maximum Capacity
Collector Size	Number of VMs (in thousands)	Flows per Day (in millions)	Flow count in 4 days (in millions)	Number of Edges for VMware SD-WAN (in thousands)
Medium	4K	2.5M	3.25M	4K
Large	10K	5M	6.5M	6K
Extra Large	35K	10M	13M	10K

Note:

The count of VMs and edges mentioned in the table is the maximum individual limit for a single deployment. So, if you have edges in your setup, you might have to reduce the VM count.
The count of VMs includes the templates on the vCenter as well.
For a single deployment with more than one collector, the limitation on the total flows across collectors is based on the capacity of the platform.

The maximum time skew between the platform nodes has to be lesser than 30 seconds.
The availability of the NTP service is critical to system operations. Ensure that you do not reboot the platform node or the collector node when the NTP service is not available.
When the existing compute resources are completely used by the other processes on the platform, vRealize Network Insight crashes and does not recover automatically. If the services fail to recover, reboot the platform node.
If the network latency between platform node and upgrade server is greater than 500ms, the vRealize Network Insight upgrade might encounter an error. So, the network latency must be less than 500ms.
The recommended disk latency for optimal performance is up to 5ms. If the disk latency is greater than 5ms, the system performance degrades.
The recommended disk IOPS is 7500.

You can customize vSphere HA options to enable vSphere high availability.