Say I have 2 identical unmanaged gigabit switches on a network, A and B. There are computers connected to switch A, there are other computers connected to switch B. A and B are connected via 1 cable. So, the total (theoretical) bandwidth available for computers on A to talk to computers on B is 1Gbit/sec. All is well.

If I now connect a second cable between the switches, and make no other changes, would the total (theoretical) bandwidth between the 2 groups of computers go up to 2Gbit/sec?



  • If your unmanaged switches are so simple that they don't do Spanning Tree Protocol (STP), connecting the second cable between them would cause a bridging loop, instantly saturating your network, making your network unusable until you remove the second cable.

  • If they do do STP, then only one of the switch-to-switch links will be used.

  • If you had manageable switches instead, then you could configure them to aggregate/trunk multiple ports together, getting a speed increase slightly less than the sum of the aggregated ports' speeds (there is some overhead from aggregation).

  • If you get less than the theoretical bandwidth through an aggregated link it is not really because of aggregation overhead. Rather it is because flows may not be spread completely evenly across the links, and if one of the links happens to get flows that aren't capable of filling the link then your trunk isn't fully utilized even if another link in the trunk may be saturated. The extreme case of this would be when you have only 1 flow to be sent through the trunk (or 2 flows that by chance gets mapped to the same link). – kasperd Sep 19 '17 at 6:10

Not unless you configure link aggregation between the switches. Depending on your switch, if it even supports it, this may be referred to as NIC Teaming, Etherchannel, Port Channel, or trunking.

Note that if you simply plug in two connections without configuration such as this, you could seriously degrade your network by causing a loop, as @Spiff mentions in his post.


Could be. I could see the switches connecting A1 to B1 through one path, and A2 to B2 through another (A1 and A2 are next to A, B1 and B2 to B). The plain spanning tree protocol will always pick the same link, however.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy