There are two scenarios where I’ve found traffic shaping advantageous. The first involved an Ethernet handoff with a higher-speed transport such as an OC3, and the second involved Frame Relay with a large number of nodes. Let’s start with the former.
Telco providers now offer Ethernet handoffs from traditional circuits. This can have the significant advantage of reducing cost for the end customer. Take, for example, an OC3 circuit. To terminate an OC3, you’ll need a router that can support an OC3 card. These routers and the requisite cards are a bit more expensive than your run-of-the-mill remote office routers. If you can get that same OC3 circuit, converted by telco into a Gigabit Ethernet connection, you could plug that link directly into one of your existing switches.
There are some downsides to this that may not be immediately apparent. The first is QoS. If you require LLQ on your link, you may not be able to provision it on a switch in the way I’ve shown. Remember, switches are designed for LANs, so they generally employ CoS, not QoS. As such, many switches will simply not accept the commands I’ve shown previously in this chapter, and you won’t be able to apply your QoS scheme in this way. If this is a requirement, plan accordingly up front.
I’ve made this mistake so you won’t have to. We were planning a nationwide rollout of MPLS and ordered Ethernet handoffs with the assumption that we could just plug into existing switches. ...