It has been always an interesting question for us when asked during the Partner Enablement session. Standard Definition about each Virtual Switch could be found in VMware documentation but it’s scattered. Let’s bring that information together:
What is Virtual Switch or vSwitch?
Virtual Switch is similar to Physical Switch but abstracted from the Hardware. Similar to Physical switch it has multiple ports, each of which can be connected to a single machine or switch on the network. Virtual machines have network adapters (vNICs) that connect to the port groups on the switch and connect uplink ports on the switch to the physical NICs of the hosts so that Virtual Machines can communicate to the external network via those Uplink ports.
vSphere Standard Switch or vSS
vSS is a default virtual switch that is installed on the ESXi host and configured with management kernel port. vSS provides bridging of the traffic internally between virtual machines inside the same VLAN and also link to external networks via uplink ports to provide network connectivity. Each ESXi host has independent vSS that has to be managed individually on each HOST.
While building data center server inventory commonly those physical server configurations are matched for Day-2 operation ease. For management and troubleshooting proficiency it was also common to have a homogeneous teaming policy and switch configuration on every host configured inside the cluster or Data Centre. It was an administrative overhead and cumbersome task to manage vSS individually on each host so VDS came out as an enhanced version of virtual switch that scales better than vSS.
Virtual Distributed Switch or VDS
As depicted VDS is the Enhanced version of virtual switch that can be configured and centrally managed from vCenter server. VDS is distributed in nature which allows you to configure it once and then simply add it to ESXi host (Which inherits setting associated to that switch). Any networking configuration change (In association to VDS) that you perform on vCenter Server is automatically pushed down to all host. There are much enhancement in VDS over vSS but one which is very important from networking standpoint is NIOC providing QOS during bandwidth contention over converged network adapter.
NSXv is marketed and mandates the use of vSphere distributed virtual switch (VDS) in vSphere for network virtualization and on other hand vSS was not supported. NSXv Manager is mapped to single vCenter to support Network Virtualisation, single VDS for overlay and teaming policy had to consistent throughout the Cluster prepared for NSX . Also Transport Zones are bind to cluster in vCenter defining the span of Overlay Logical Switch inside the Data Centre.
While NSXv is leading the market and VMware vision was to connect and protect any device, application, and Cloud. NSX-T as a platform (Advent of NSX-MH) supported multi hypervisors, bare-metal Server, containers and Native workloads in Cloud. And to support networking functionality on all those it was necessary to use a construct that was not bound to vSphere, hence N-VDS was developed to support such a versatile environment. Now this brings us to the next form of virtual switch which N-VDS
NSX-T Virtual Distributed Switch or N-VDS
N-VDS is not very different than Virtual Distributed Switch that owns minimum two or more physical NICs in the production environment. N-VDS can only be configured and managed within NSX-T Manager and not via vCenter contrary to VDS or vSS. This decoupling from vCenter allows us to use N-VDS outside of vSphere environment.
In NSX-T Transport Zone can span endpoints across different platforms or multiple vSphere vCenter hence N-VDS is correlated to a transport zone and created when the transport zone is created. NSX Switches (N-VDS) of the different transport nodes are independent but can be grouped by assigning the same name for centralized management.