Hubs, Managed and Unmanaged Switches

hubs, managed and unmanaged switches

When I was very, very young, I remember my grandmother having a party line telephone. Before AT&T could build out the phone system, it was easier to connect groups of families to one line than supply each subscriber with their own phone line. In my grandmother’s case, I’m sure she had a party line because it was cheaper than having her own phone line.

It’s hard to believe now but you not only shared a phone line with the people in your house but with a group of other nearby families – people that you often didn’t know. To make a call, you picked up the phone and listened to see if others were using it. If they didn’t hear you pick up the phone, you could just eavesdrop on their conversation or you could provide your unsolicited thoughts on their conversation. If you had to make a call, you would have to try to get them to end their call. It’s an interesting time in the phone system history.

A hub is nothing more than that old-fashioned party telephone line. A hub works exactly like a party-line telephone. Any device on any port of a hub gets any and all traffic from every other port. To send a message, a device connected to a hub starts sending and then monitors the line to see if anyone else is sending. If two devices are sending at the same time, the messages collide, and the senders stop and resend their messages later. It’s just as inefficient as waiting for others to end their call to make your call.

Hubs are rarely used in industrial automation applications. They’re exceptionally cheap, so if you had devices that rarely transmitted information, it might be a good way to architect that part of your system. If you had some sensors that sent data rarely or even once a second, a hub would be a good, cheap way to connect them. If messages did collide, there would be plenty of time (bandwidth) for those messages to be retransmitted.

If you have more important devices that need consistent connections, like EtherNet/IP Adapters or Modbus TCP Slave devices, you could use an unmanaged switch. Unmanaged switches are a bit more sophisticated than the party line hubs. Unmanaged switches operate at the Data Link layer (Layer 2) of the OSI model. They build an internal table with the 48-bit Mac addresses of all the devices on the switch. Whenever a message arrives on a port, it looks at the Mac address to identify what port it should send the message to. For messages that must be delivered to some distant Ethernet address on the other side of the plant or on the internet, one of those ports is connected to a router that understands how to route messages through networks.

Managed switches are unmanaged switches on steroids. It’s the difference between riding in the family car and a souped-up limousine on prom night. Unlike unmanaged switches, managed switches are Ethernet devices in themselves. Some are even EtherNet/IP devices. They have an IP address, a Mac address, and a web browser interface. Managed switches provide the ability to configure, manage and monitor the network of devices attached to them. They have all sorts of advanced features:

  • A web browser for configuration
  • Diagnostics that tell you what is happening on the network
  • SMTP interface for remote control and monitoring of your switch
  • Ability to segment your network into different broadcast domains for security or performance
  • Quality of Service to make sure high priority messages are delivered on time
  • Port Mirroring so that you can monitor traffic on one port from another port

Communications has come a long way since my grandmother’s party line telephone – from party line telephone to managed switches.

I’m excited to see what’s next.