So we all know that Cisco is big on IP and Ethernet Multicast in their certification programs. In real life, we don’t actually get to see a lot of use for Multicast. In the main, the the main application for Multicast would be for watching Television, but the current consumer doesn’t want to watch content at a ‘programmed’ time, they want to stream content according to their own timetable.
BBC iPlayer
Here in the UK, the BBC streams most television shows for seven to fourteen days after they have aired using their iPlayer HTTP streaming client. This means that you can sit down at anytime and watch a particular show when you want. This service has been hugely popular (the sound of Service Providers whining about bandwidth is quite loud) and gives a good guide to future of the Media Content as a unicast data source.
If the only people using Multicast are financial trading applications, and Cable Companies in just a few countries, why are we bothering ? I don’t have any good answers.
Why is Cisco into it ?
Cisco LOVES Multicast because it sells bigger routers and switches. Most importantly, it makes them a key supplier in any debate / discussion / plan at the CIO level in organisations. Networking is often ignored because it’s not visible at CIO level and getting exposure means that you can control / participate the sales cycle for upcoming projects.
So we suffer
I’m speculating that Cisco makes sure that Multicast is in the curriculum to ensure that any customer deciding to use Multicast as a key technology can feel confident that engineers have the knowledge.
And many people suffer with learning irrelevant material. And so the IETF continues to push out Multicast protocols. If anyone know why we bother with this, except as an academic exercise, please get in contact or email myetherealmind — at __ gmail-com, I’d love to find out more.
RFC 5790 – Lightweight Internet Group Management Protocol Version 3 (IGMPv3) and Multicast Listener Discovery Version 2 (MLDv2) Protocols: ” Lightweight Internet Group Management Protocol Version 3 (IGMPv3) and Multicast Listener Discovery Version 2 (MLDv2) Protocols”
Loading ...
About Greg Ferro
While I get it multicast seems like overhead, it works and has for a long time. There are new protocols because things change and so do the requirements. Mutlicast for VoIP, Video, corporate training, financial applications, the list goes on.
The comments just reminds me of a Cisco employee I spoke with a long time ago who said EIGRP is perfect who needs anything else. Reality is lots of people, and not every block fits every hole. If you were to knock on the door of 1000 business in NYC or San Fran or Chicago or London or Tokyo, at least 250 use multicast regularly and probably another 250 do but don’t really know it.
There’s plenty of applications that can utilise multicast to be more efficient and reduce burden to the network. Too many people get caught up in the idea of consumers and video on demand etc. One example you may not have thought of: stock ticking. It’s a (relatively) low volume one way feed that people consume but don’t get involved with otherwise. Yet if you have hundreds of thousands of customers each wanting the same feed the infrastructure cost to support a unique feed to each is quite high even if you’re using UDP and not TCP.
You say this because you work private sector. Public sector is big into up-to-the-minute news and Multicast perfectly delivers that. Using 4500s as building distribution switches, it works pretty damn good. Stream LIVE (yes, many people want live, not short segments) CNN on a few big screens and people are happy.