Hey Nerds, Geeks, and Ziglets out there. Today we start our Mini Series on Segment Routing with our Guest Expert Suraj Soni. We have a lot planned for this Mini Series but today's focus is "Why Segment Routing?" so lets get to it!
Guest Expert: Suraj Soni
Soni has been working in our industry for 10 years in multiple verticals to include Enterprise and Service Provider Networks. He is a triple CCIE in R&S, Sec, and Service Provider. He has numerous other Vendor Certifications such as F5 and Nokia. He has been in a number of different roles in his career such as Solution Architect and Network Consultant. He currently is a Senior Network Manager for a company in Singapore. Please Welcome my friend Soni to our show today!! Welcome Soni!
Soni joins the Zigbits Network Design Podcast today to start a Mini Series on Segment Routing! We are going to have about 5 podcast episodes and a number of blog posts to help reenforce the topics we are covering. Today's topic is "Why Segment Routing"?
Why Segment Routing?
Segment Routing is a control plan protocol for MPLS. Segment Routing came into picture because of the evolution of Software Defined Network Solutions (SDN) and the desire to expand said SDN solutions and concepts into the Service Provider arena.
Why is LDP and RSVP out, why do we need SR? One reason is because with LDP and RSVP we are statically mapping our policy to an IP address. In addition with LDP and RSVP we have no visibility of the applications running over the Service Provider Core without implementing some other technology or solution.
Sample Use Cases:
Voice and Video - Needs low latency
Higher bandwidth needed for file download
In both of these use cases, with LDP and RSVP we have static uni-directional configurations of MPLS TE Tunnels which in most cases are never removed even when they are no longer needed. Lets simplify this to say unnecessary TE Tunnels and complexity.
How Segment Routing different:
For Segment Routing there is no signally like there is with LDP, RSVP, and MPLS TE. Segment Routing is a signal free protocol because it instead uses your IGP for signaling. The IGP floods the Segment Routing information as well when it sends out other information updates. Currently only OSPF and IS-IS are supported.
Segment Routing is not going to resolve these Use Case issues by itself. Segment Routing is a part of a solution.
A Solution can resolve issues, a technology cannot resolve an issue.
Open Controller (SDN)
Within this entire solution we need a brain or an intelligence center to give us some of the SDN pieces we desire in the Service Provider Core. This is where we bring in a controller of some sort. This controller is going to have both the visibility of the customer locations (Data Center, WAN, Campus, etc..) and also the Service Provider Core network. This allows the controller to have a full End to End visibility of the network. With the full End to End visibility, the controller is seeing the use cases happen in real time.
The controller and application will have a form of integration which we will show in a later segment of this mini series. This integration will give the controller the ability to fetch the details for the applications in question. Then the controller will find the path throughout the End to End network dynamically keeping measurement of all of the necessary health controls of the different paths and the application requirements (think latency and jitter for voice here as an example). Once the path is identified the controller tells the application what the path is by giving the application packets the Segment ID list.
With this solution, nothing is configured on the routers like MPLS TE or RSVP. The only requirement on the routers is to properly configure Segment Routing.
What if I don't have a controller, you can still use Segment Routing, but you will rely on label forwarding, more to come in future
Segment Routing Summary