Orhan Ergun No Comments

Today, there are many networking technologies which haven’t been widely deployed. And among them are Internet Multicast and IPv6 although these two protocols have many benefits . 

But probably people are asking the correct question. Do we really need new protocol ? Or can we solve our problem with the existing mechanisms deployed on our network ?

These two questions are the most important questions which every network designer should ask in every new deployment.

Findings show that even IPv4 exhaustion couldn’t help IPv6 deployment much.People have considered either to deploy CGN (Carrier Grade Nat) or to focus on the IPv4 transfer markets.

Like IPv4, Internet Multicast is not alone in this. Internet Multicast could help distribute content on the Internet very efficiently. But due to many deployment problems and lack of interest, CDNs (Content Delivery Networks) used unicast delivery rather than utilizing Internet Multicast.

Although LISP (Locator Identity Separation Protocol) seems very promising and could solve many networking problems, it was unfortunately just another encap/decap protocol which uses an authoritative server for the endpoint reachability.

Moreover, many use cases of LISP could be done with the existing technologies. An example is the inter-domain routing policy. We use BGP as the inter-domain routing protocol on the Internet. Although ingress load balancing may not be efficient with BGP,  by playing with communities, AS path attribute and specific prefix announcement, ingress load balancing problem are solved.

Granted, the result may not be perfect but probably good enough so that people don’t look at the immediate alternative. They have tried to solve the problem with the existing tool on their network. 

In addition to that, many of these protocols require special hardware or software support which creates operational expenses. And due to the size of large-scale networks, upgrade usually takes much time, and the requirements cannot be fulfilled quickly.

I know a couple of networks which deployed LISP and operated them successfully. But out of hundreds of networks which I was involved in, this number is definitely insignificant. And although LISP has many use cases, almost all of them can be solved with the existing mechanisms and technologies.

Let me know what do you think about LISP deployments, share your thoughts in the comment box below.

That said, it is probably time to consider AMT (Automatic Multicast Tunneling) and discuss whether it could solve Internet Multicast problem. And who knows, perhaps it could be an alternative to CDN.

Leave a Reply

Your email address will not be published.