Traffic Engineering · Topic 1 of 3
Traffic Engineering Basics
Learn what MPLS traffic engineering is, why shortest path routing wastes capacity, and how a TE tunnel steers traffic over a chosen path on RouterOS.
- 4 min
- Traffic engineering steers traffic over chosen paths to use capacity well.
- Shortest path routing can overload one path while others sit idle.
- A TE tunnel is a label switched path placed on a chosen route.
What it is
Traffic engineering is the practice of steering traffic over chosen paths so that network capacity is used well.
Ordinary routing sends all traffic along the shortest path. That path can become full while a longer path next to it sits idle.
MPLS traffic engineering solves this with TE tunnels. A tunnel is a label switched path that you can place on a path of your choice.
How it works
A tunnel is defined at its head end, a path is found, a signalling protocol sets it up, and traffic is steered into it.
The problem with shortest path
An interior protocol like OSPF picks the lowest cost route. All flows to a destination follow that route, and the other links are unused. A link can be saturated while the alternative has plenty of room.
Changing link costs can move traffic, but it moves all flows at once and may only shift the problem elsewhere. Traffic engineering moves selected flows without changing the routing table of the whole network.
Parts of a TE system
| Part | Job |
|---|---|
| TE tunnel | A one way label switched path from a head end to a tail end |
| Path | The list of routers the tunnel should follow |
| Signalling | RSVP sets up the path and distributes the labels |
| Link information | Available bandwidth on each link, shared by the interior protocol |
| Steering | Rules that send traffic into the tunnel |
A TE tunnel is unidirectional. Traffic in the other direction needs a second tunnel configured on the other router.
Tunnel configuration on RouterOS
MPLS must already be running. A TE tunnel is an interface that points at a tail end address.
/mpls traffic-eng tunnel-path add name=path1 use-cspf=no hops=10.0.1.2,10.0.2.2
/interface traffic-eng add name=te1 to-address=10.255.255.3 primary-path=path1
/interface traffic-eng printtunnel-pathdescribes a route through the network.interface traffic-engcreates the tunnel interface.to-addressis the tail end.primary-pathselects the path the tunnel should use.
The details of path types and bandwidth belong to the next two pages. The main point here is that the tunnel is a normal interface in the router.
Steering traffic into the tunnel
A tunnel that exists carries nothing until traffic is directed to it. One simple method is a static route that uses the tunnel interface as its gateway.
/ip route add dst-address=172.16.0.0/16 gateway=te1The packets for that destination enter the tunnel, receive a label, and follow the chosen path. Other destinations continue on the normal shortest path.
What a tunnel does not do
A TE tunnel does not change routing in the network. The routers in the middle do not know about customer or destination prefixes. They forward by label, and only the head end decides which traffic enters.
Why use TE
- To avoid congestion by spreading traffic over several paths.
- To place important traffic on a better path.
- To reserve bandwidth along a path, which a later page covers.
- To provide a backup path that is already planned.
Remember the three parts of any TE design. Define the path, set up the tunnel, then steer the traffic into it.
What to remember for the exam
Seven facts cover most questions here.
- Traffic engineering steers traffic over chosen paths to use capacity well.
- Shortest path routing can overload one path while others sit idle.
- A TE tunnel is a label switched path placed on a chosen route.
- A TE tunnel is unidirectional, so the return direction needs its own tunnel.
- On RouterOS the tunnel is an interface created with
/interface traffic-eng. - Paths are described in
/mpls traffic-eng tunnel-path. - Traffic must be steered into the tunnel, for example with a static route.
A TE tunnel that is up does not carry traffic by itself. Without a route or policy that uses it, the packets keep taking the shortest path.
Common mistakes
These four errors are common.
Expecting the tunnel to carry traffic automatically. It only does so when a route or policy points into it.
Configuring one direction only. Return traffic still follows the normal path.
Starting without working MPLS. A tunnel needs a labelled path and reachable loopbacks.
Believing TE changes the routing table of the core. Only the head end decides which traffic uses the tunnel.
Check yourself
Work through these three questions.
Why is shortest path routing not always efficient?
All traffic follows the lowest cost path, so it can become full while a longer path sits idle.
Is a TE tunnel bidirectional?
No. It is unidirectional, so the other direction needs its own tunnel on the other router.
What does a TE tunnel need before it carries traffic?
A route or policy must send traffic into it, for example a static route with the tunnel interface as the gateway.
Topic 1 of 3 in Traffic Engineering
Test yourself on Traffic Engineering
The full MikroTik MTCINE question bank has 475 questions, with timed practice exams and an explanation for every answer.
This study guide is independent and is not affiliated with, endorsed by, or sponsored by MikroTik. Vendor names, certification names and exam codes are trademarks of their respective owners and are used for identification only.