Skip to content

VKS Network Services

This section describes the procedures for deploying the Supervisor within a vSphere environment.

Different network options for Supervisor deployment

Explore the specific network architectures available for your Supervisor cluster:

Architecture
Description & Use Case
1. VDS + FLB VDS Architecture

  • Architecture: Uses VDS port groups and deploys FLB virtual appliances to handle load balancing traffic.
  • Best for: VMware vSphere Foundation (VVF), lab environments, Proof of Concepts (PoC), or smaller environments.
  • Limitation: Limited Network Services (only LB) and limited VCF Automation support (no namespace create/delete/update from VCFA).
2. NSX + DTGW
or NSX + CTGW		 NSX Architecture

  • Architecture: Uses NSX Distributed or Centralized Transit Gateways and VNA or Edge Cluster to handle load balancing traffic.
  • Best for: Fully integrated VCF architecture for better scale and security.
  • Consideration: Requires fully deployed NSX overlay infrastructure with DTGW+VNA or CTGW+Edge+T0.
Detailed Architecture Pros & Cons

1. Supervisor with "VDS + FLB"

  • Pros:
    • Footprint: Slightly smaller overall footprint (the FLB appliance is slightly smaller than a VNA)
    • VMware Editions: Available in VMware vSphere Foundation (VVF) and VMware Cloud Foundation (VCF)
  • Cons:
    • Network Services Limitations: Lacks full support of Network Services such as Subnets, Static Routes, NAT
    • VCF Integration Limitations: Lacks full support for other VCF components, specifically VCF Automation (VCF-A)
    • Scale: All VIPs are managed by a single FLB's Active/Standby (A/S)
  • Requirements:
    • L2 Connectivity: L2/VLAN connectivity required across vCenter cluster(s)

2. Supervisor with "NSX + DTGW" or "NSX + CTGW"

  • Pros:
    • Security:
      • No possible external access directly to K8s Nodes
      • Option to control application communication including:
        • Cross-Container via Antrea CNI (DFW)
        • Cross-VPC VIP communication (VPC Connectivity Policies)
    • Scale:
      • Uses fewer public IPs (K8s nodes use private IPs)
      • VIPs are highly scalable, distributed across up to 10 VNA Nodes in an Active/Active (A/A) configuration
  • Cons:
    • VMware Editions: Only available in VMware Cloud Foundation (VCF) (not VMware vSphere Foundation (VVF))
    • Operations: Requires NSX (though installation and management remain very simple)
    • Footprint: Slightly larger footprint (the VNA is slightly larger than the FLB)
  • Requirements:
    • For both DTGW/CTGW: Large MTU (min 1700 - recommended 9000)
    • For DTGW: L2/VLAN connectivity required across vCenter cluster(s)
    • For CTGW: Physical network with BGP routing protocol
Upcoming Architectures

The following architectures will be covered in future updates:

  • Avi Load Balancer

Document Versioning

This guide is updated for VCF 9.1+.
If you are running an older version, some options may not be available.