In the beginning when the Internet was first being developed there was a metaphor developed by Dave Clark at MIT which was that the Internet was an hourglass, thin in the middle with all of the intelligence at the edges. This maximized innovation. It worked for the past 40 years.
We show this below. The intent was to keep the middle, the IP stuff and below, easily changeable because we all knew that no one had the best ideas and we would need to continually update the net. Moreover we knew that for those of us who had developed large switches, where the intelligence was in the network and was hierarchical that change was impossible. Thus keep the intelligence at the edge, never let it migrate to the center.
Now we have the classic protocol stack, albeit now a bit aged but it does present a reasonable paradigm. Thus we see layers at the bottom, 1 thru 3, stay within the net, whereas all others are user based, or should be.
We first look below at the Cisco/AT&T Telepresence architecture. It is a true throwback to the past. It demands an internal centralized controller ostensibly for scheduling. My current video systems do not need such a centralized approach and they work peer to peer. The create multimedia communications paths using session layer. In 1987 I wrote a paper detailing this for broadband and it is slowly being adopted.
Now below we show the hourglass approach. We use the outer layers as the control layers and this will allow any user set to communicate with any other user set.
Now we have several issues resulting from an analysis as above.
1. Net Neutrality is a true blurring of this model of the hourglass. It starts to put things into the center of the network for such euphemistic purposes such as network management. Nonsense, network management in the simplest sense is an SMTP protocol on IP. It does not belong inside the network.
2. Video to thrive must be built on an hourglass. The best will prosper in a true Darwinian sense. Doing say what Cisco is doing in Telepresence is a true throw back to what is not a well know dated artifact of the past, centralized control. The presence of AT&T in this mess is a true sign of its pending collapse. The use of say Internet 2 rather than the AT&T system is a true step forward.
3. Video will and is becoming the largest growing user of the bandwidth. This will be true on both wired and wireless applications. One need look not further than LTE architectures to see its presence. Thus it is essential to keep the edge alive and growing to enhance the video apps.
We show this below. The intent was to keep the middle, the IP stuff and below, easily changeable because we all knew that no one had the best ideas and we would need to continually update the net. Moreover we knew that for those of us who had developed large switches, where the intelligence was in the network and was hierarchical that change was impossible. Thus keep the intelligence at the edge, never let it migrate to the center.
Now we have the classic protocol stack, albeit now a bit aged but it does present a reasonable paradigm. Thus we see layers at the bottom, 1 thru 3, stay within the net, whereas all others are user based, or should be.
We first look below at the Cisco/AT&T Telepresence architecture. It is a true throwback to the past. It demands an internal centralized controller ostensibly for scheduling. My current video systems do not need such a centralized approach and they work peer to peer. The create multimedia communications paths using session layer. In 1987 I wrote a paper detailing this for broadband and it is slowly being adopted.
Now below we show the hourglass approach. We use the outer layers as the control layers and this will allow any user set to communicate with any other user set.
Now we have several issues resulting from an analysis as above.
1. Net Neutrality is a true blurring of this model of the hourglass. It starts to put things into the center of the network for such euphemistic purposes such as network management. Nonsense, network management in the simplest sense is an SMTP protocol on IP. It does not belong inside the network.
2. Video to thrive must be built on an hourglass. The best will prosper in a true Darwinian sense. Doing say what Cisco is doing in Telepresence is a true throw back to what is not a well know dated artifact of the past, centralized control. The presence of AT&T in this mess is a true sign of its pending collapse. The use of say Internet 2 rather than the AT&T system is a true step forward.
3. Video will and is becoming the largest growing user of the bandwidth. This will be true on both wired and wireless applications. One need look not further than LTE architectures to see its presence. Thus it is essential to keep the edge alive and growing to enhance the video apps.
Posts
