Managed Switches and 5-4-3-2-1 Policy

Unanswered Question
Oct 19th, 2007

I was reading over a power point presentation and found the 5-4-3-2-1 rule:

The 5-4-3-2-1 rule requires that the following guidelines should not be exceeded:

Five segments of network media

Four repeaters or hubs

Three host segments of the network

Two link sections with no hosts

One large collision domain

What I was curious about is if managed switches would be part of that rule? I'm working with a few other engineers on designing a new Ethernet/IP control network and I haven't been able to find much information. Each port of the switch will have a multicasting device sending out information in timed intervals. We're doing a ring topology using STP. Thanks in advance for any help.

I have this problem too.
0 votes
  • 1
  • 2
  • 3
  • 4
  • 5
Overall Rating: 0 (0 ratings)
konigl Fri, 10/19/2007 - 22:33

The 5-4-3-2-1 rule is for 10 Mbps half-duplex Ethernet. This rule defines the design parameters for establishing a single collision domain.

On a switch, each interface can support a collision domain. Multiple interfaces on a switch means there are multiple collision domains attached to that switch. A switch's job is to bridge network traffic between or among those multiple collision domains.

The 5-4-3-2-1 rule doesn't apply to switches. It applies to the equipment you attach to switches, if that equipment happens to be running 10 Mbps half duplex Ethernet.

There are different rules for 100 Mbps half duplex Fast Ethernet. I won't get into the details here, but they are more restrictive than the 5-4-3-2-1 rule.

Full duplex operation at 10/100/1000 Mbps speeds is generally used on point-to-point connections: switch-to-switch, server-to-switch, or workstation-to-switch. Cable distances are pretty much the only thing you need to worry about there.

As you interconnect Layer 2 switches with full duplex links, though, you need to pay attention to the topology you build. Especially when you start introducing redundant connections that create loops in the switch topology. You need to run STP to keep the loops from making your network unusable due to excessive broadcast traffic.

STP's timing defaults are based on a Network Diameter of 7. That is, to get from one edge node on a switch to any other edge node on any other switch, worst-case scenario is you should cross at most 7 or less switches along the way. Any more than that, and STP may not be able to do its job correctly.

A ring of 7 switches would not be a problem. A bigger ring of 21 switches, could be a big problem.

Now, you CAN adjust the timers if your Network Diameter is greater than 7. But it would be a better idea to re-design your network to "flatten" it out, cutting down on the maximum number of switch hops between any two nodes.

In the 21-switch scenario above, maybe a better way to do it would be to create three 7-switch rings. And use routers or Layer 3 switches to join those rings together.


This Discussion