VLAN Structure

A LAN consists of stations, repeating hubs and switching hubs operating at the data link layer. LANs could be connected to other LANs if routers are used; thereby, creating an internetwork. Each LAN would then be given a network address. The best example of an internetwork is the Internet. Therefore, it is possible to have the industrial automation system on one LAN and the information system on another LAN with the two linked by a router. However, the structured wiring within the plant may not support this wiring directly. Besides, configuring routers is more difficult than configuring VLANs. What is desired is to have the information system and industrial automation system on the same LAN, but logically separated into two LANs. That is what a VLAN can do.

Within a LAN that has all stations connected to repeating hubs, all stations hear all three types of transmissions—unicast, multicast and broadcast. In this situation, it is not possible to establish separate VLANs since there is no way of restricting traffic. A basic requirement of VLANs is the use of switching hubs. A switch learns the location of stations by observing the source MAC address present in a message received at an incoming port. The MAC address-port number association is so noted in its filtering database. All future transmissions destined to a MAC address that is stored in the switch's filtering database, will only be directed to the port associated with that MAC address unless the transmission originated on that port. If a MAC address is received with no association, the transmission is flooded to all ports (except for the received port) as if the switch were a repeating hub. The same is true for multicast and broadcast messages. Therefore, a switch provides an improvement in performance over repeating hubs by restricting unicast messages to only those stations involved, but it is this filtering capability that can be exploited for VLAN use. A single switching hub can be so configured and thus act as several independent switching hubs by creating VLAN associations to switch ports.


There are several ways of creating VLANs, but the easiest to understand is the Port VLAN. Switches create an association of MAC addresses and port numbers. What needs to be added is a VLAN association. This would have to be accomplished through some configuration of a switch that can support VLANs. VLAN support is not possible with a Plug and Play switch — one with no means of altering its personality through operator intervention. For example, within a sixteen-port switch we want to create three separate VLANs numbered one to three. During configuration, we associate each port on the switch with a VLAN. From then on, traffic within a VLAN assignment will be restricted to only those ports associated with that VLAN assignment. Using our example of three VLANs, we established VLAN1 as associated with ports 1, 2, 3 and 4. A broadcast or multicast message on port 1 would be sent only to ports 2, 3, and 4 and no others. The other VLANs would operate in a similar fashion. A unicast message would be forwarded as with any other switch. There would be a MAC address-port number association. However, added to this association would be the VLAN constraints. So if the MAC address-port number association is not present in memory for a destination address, flooding will only occur with the VLAN port group. What happens when a destination address is specified in a transmission received on a port from another VLAN group? The transmission should be discarded.

Figure 2 shows a Port VLAN application consisting of three VLANs, although more VLANs can be added. There is only one VLAN-aware switch located in the middle of the LAN. The other switches that are not VLAN-aware are considered part of the respective VLANs. Each port on the VLAN-aware switch has an association with a common port on the switch where a server resides. This overlapping of VLANs allows any workstation in a VLAN to access the server, but workstations in separate VLANs are not known to each other.

Figure 2 — In this Port VLAN application, the server in the middle is logically attached to all three VLANs.

A big advantage of Port VLAN is that it is simple to understand and use. Patch panel ports can be tagged with the associated VLAN, and it is just a simple matter of moving patch cords around to connect particular stations to particular VLANs. A simpler way of doing it is to have software do it. By reconfiguring the VLAN-aware switch, physical ports can be reassigned to different VLANs. However, what if you want to stretch your VLAN across several switches? It is possible, but you would need to have dedicated wiring for each VLAN. That is a severe restriction and, therefore, Port VLANs are best accomplished using a single VLAN-aware switch. Notice that there is no change in Ethernet frames with Port VLAN partitioning. End stations are unaware of the VLAN structure. More flexibility is gained if VLAN associations can be learned from the contents of the Ethernet frame. This is called implied tagging which allows VLANs to span multiple switches using the same cabling structure.

Frame-Encoded VLAN Schemes

With Port VLAN, there is no altering of Ethernet frames or any implicit tagging within Ethernet frames. Stations are unaware of the VLAN structure. There are alternate ways of establishing VLANs if the switches being used support the various schemes. You could simply associate particular MAC addresses to a VLAN. In this way the station assigned to the VLAN can be on any switch port and still be attached to a particular VLAN. Of course, if that station were ever replaced, all switches would need to be reconfigured for the new MAC address. Another approach to VLANs is to separate stations according to the network operating system being supported. By examining some protocol field, frames could be directed only to those stations supporting that operating system. This approach to VLANs was popular when there were several competing network operating systems with much different Ethernet frame definitions. The movement towards universal TCP/IP acceptance has now limited the frame structure choices. Another scheme is to define a proprietary protocol by coding the Ethernet frame with VLAN information. The problem with proprietary schemes is that they do not have wide industry support. To obtain wide industry support, you need an IEEE standard.