SR Policy | In this section, we will understand SR policy in depth as this is the main concept in Segment Routing traffic engineering. This is part 2 in SRTE series .
SR Policy defines two basic actions of SR traffic engineering ,
It express a path through the network which is different from shortest path computed by IGP
Let me remind here, SR policy is identified by three components and these are,
color – it is used to distinguish multiple SR policies on the same headend
SR Policy has one or more candidate paths , dynamically calculated or explicitly configured. Dynamically calculated path may have certain SLA objective like low delay path provided by headend node. As SR policy can have more than one path, the one with highest preference becomes active candidate path.
Let’s understand Segment Routing traffic engineering with below example , refer below illustration in figure 1,
Lets say, SLA intent is to send traffic through low delay path in the above sample network from Node 1 to Node 4. Now, headend node has to translate intent into SID list . First, headend node needs to know latency of each link where IGP comes into the picture. IGP while exchanging link state information , also exchange other link attributes such as delay of the link, loss etc. Headend node receives all these information and store it in SR-TE database.
To compute low delay path , headend node calculates cumulative delay by adding delay of each link from Node1 to Node4. The resulting path will be, S1-S2-S3-S4 as this path is having lowest cumulative delay (9+6+6=21ms). Basis this calculation, Node1 encodes path with SID list as S1-S2-S3-S4 (assume S1,S2…. etc are SIDs of respective nodes).
Now, let us defines SR policy ( as mentioned SRTE is nothing but SR policy in Segment routing, there is no concept of tunnel in SRTE).
SR policy will look like,
SR Policy (Node1, 100, Node4) where,
Node1 – Headend node
100- color (it can be any number to uniquely identify SR policy or intent or SLA)
Node 4 – Endpoint
And candidate path will be, S1-S2-S3-S4 (stack of SID list) which is delay optimised path in the network . Anytime if there is change in the network, headend node will update SID list accordingly.
I hope with the help of above example, SRTE or basic principle of SR policy is understood. We will discuss SRTE in next few articles as well with more complex examples.