There are several choices involved in setting up a vSphere virtual infrastructure, including choosing the appropriate switch type -- standard or distributed -- for your environment. Switches funnel data toward its intended destination, and a virtual switch inspects data packets to then intelligently direct communication on the network.
The key difference between standard and distributed switches lies in their scope. A standard switch works within one physical host only; it's not possible to add any flexibility or share the switch with other hosts. VMware's vSphere distributed switch (vDS) is shared between hosts, available as a virtual device between the machines involved. It functions as one single virtual switch that connects hosts as well as virtual machines (VMs).
VMware distributed switches debuted with vSphere 4 and received major upgrades in 2012. There are many similarities between standard and distributed switch types. Distributed virtual switches provide connectivity for VMs and the VMkernel interface that manages traffic between hosts. A distributed switch also needs physical network adapters to connect to the external physical network.
Distributed switches are more flexible than standard switches, allowing different hosts to use the switch as long as they exist within the same cluster. At first glance, this may seem to complicate the configuration, but in fact, distributed switches make adding new hosts to the cluster relatively easy.
Creating a vSphere distributed switch consists of two steps. First, create the switch from the data center context. Then, add all hosts that will use the virtual switch. When adding the virtual switch to the data center, you'll have the option to select between three different types. If you need Network I/O Control and port mirroring support, you'll need the version 5 vSphere distributed switch. If you also have vSphere 4.0 or 4.1 hosts, you'll need to select the switch type compatible with these product versions.
After creating the virtual switch, add a dvPort Group. Port groups are required to connect anything to the virtual switch, rendering the switch usable. To add dvPort Groups, select the distributed virtual switch and go through the setup wizard for the port group and associated ports.
While setting up a port group, you'll also specify the virtual local area network (VLAN) with which it will connect. If no VLANs are used, the port will be untagged, which is, not assigned to any VLAN. Typically, the VLAN type will be used, as it allows you to add a VLAN tag and use the port group in the specified VLAN. If the port group needs to be available to untagged traffic, do not select anything for the VLAN type.
Once you create the vDS, adding a host to the switch will be easy, because the host will be presented to the switch. From there, you can add additional switch features, such as traffic shaping, security and more.
One important requirement for distributed virtual switches that differs from standard virtual switch configuration is that all vSphere hosts must be in a cluster that’s managed from vCenter Server. You should also think about which uplinks you make available to the distributed virtual switch. A regular virtual switch has an uplink only to the local port on the physical hosts where it is used; however, the distributed virtual switch's uplink is also distributed. This ensures the VMware distributed switch can access the same physical network infrastructure independent of which VMware vSphere host is used.