Transmission Network and addressing network-cloud age Information Technology

I recently read the book "Information Technology in the cloud age", which is quite rebellious. It is pointed out that the age of resource abundance is approaching, and the network theory established in the Age of resource shortage must all be retried! The book sharply points out that many of the current theories are so-called "smart", or a repair, which is consistent with what I call "Tips, therefore, I found a common voice. The author believes that the network should be simple and only provide a transparent channel. All data streams are self-addressable to the destination, and there are no routers or other intermediate addressing devices that increase latency, the concept of QoS is no longer needed in the Age of resource abundance, because the concept of bandwidth is unlimited. The only problem is to increase the management plane.
Many of my ideas in this book coincide with some of my ideas. The discussion of optical networks is basically the same as my thoughts. Especially when talking about the self-addressing mechanism, it simply relies on optical components. Although the author has not explicitly stated this, it does not mean self-emergence (data goes directly to the destination through a fully established network, instead of route-based jump-by-Jump !). My disagreement with the author is that the all-established network is a transmission network rather than an addressing network, which I will discuss in detail below. In short, the author has repeatedly said that the network should not be implemented based on computers, and the network only needs to transmit data transparently. The nature of the network is simple. The reason why the current network is so complex is that, the reason is that the current network originated from the age of resource shortage.
In many sense, the author's understanding of the nature of the network is correct. For example, the pseudo header of the TCP checksum is basically meaningless, but in the packet loss rate, when the error packet rate is very high, more validation protection has to be added. In addition, the re-transmission mechanism implemented at the link layer has a large number of verification codes at each layer... these are basically unnecessary in the Age of abundant and credible resources!
Although the ideas in this book are bold, I think most of them are too empty. The following is an example:
1. OSI layered model needs to be pushed down
First, OSI is just a layered model. For the lower layer, it aims to unify the vendor interface, and for the upper layer, it aims to better design the Application Network API, as a real example, TCP/IP compresses the upper layer. The author did not propose an alternative to the OSI model. Even the so-called three-tier architecture is nothing more than an implementation method of the layered model.
2. IP router needs to be canceled
The author only says that the router delays data packets. The IP router in the OSI model is the root of evil, and also proposes a self-addressing mechanism. However, it does not specifically describe how to do this, when it comes to the all-optical network, the author only proposes that the current optical and physical devices are still in very early stages and cannot be used on a large scale.
3. In the future, only video will be available on the network.
Because the author's company is engaged in video, he can fully understand the video. However, raising its importance to an absolute height is misleading, in the big data era, many data volumes are transmitted over the network. The purpose is not only to display data on a "large-screen TV", but also to perform distributed scientific computing. A large amount of bandwidth is required!

Why do I always think that TCP/IP is very successful and I don't agree with the author's point of view in "Information Technology in the cloud era"? That's because I don't think the author has a good understanding of TCP/IP networks. The TCP/IP network actually has three layers:
Semantic layer:This layer implements end-to-end semantics, namely TCP and UDP.
Addressing layer:This layer implements logical addressing, which is the responsibility of IP.
Transport Layer:The layer is the link layer (including the OSI Physical Layer ).
If the above layers are abstract, I want to thoroughly understand the following example:
A Jun and B Jun are at the beginning of the two mountains. They can see each other. A Jun took a private package from c jun and sent it to d Jun. He wants to transfer it through B Jun, but there is a long way between the two mountains, if you can see each other, you cannot deliver the package to the other party. There are many options for Mr. A. He can hire a carriage/taxi and ask the car to deliver the package to Mr. B, he can also wait for the bus to arrive at June B. He can also ask the courier to collect it directly... in the above scenario, the package itself is the semantic layer. A Jun and B Jun know how to deliver the package and can see the other party. This is the addressing layer, while the car, bus, express delivery is the transport layer.
Through the above description, the network itself is simple. Even when we were born in a age where resources were scarce, it was still very simple. From the perspective of a Jun, B Jun, and the package, we cannot see any complexity, this is also the essence of TCP/IP, smart compression in the c jun and D Jun, as D Jun saw the content in the package is crying, laughing, dead, live, a Jun and B Jun is no matter. So where is the complexity? Complexity is added by future generations! It is not the core of TCP/IP. If we maintain the core of TCP/IP and connect the gully between June A and June B with a bridge like the San Francisco bridge, the efficiency will be greatly improved, therefore, QoS is the transmission network (as the link layer ?) !
Addressing network in a new generation of Network Independent into a signaling network, its goal is to configure all-optical transmission network, similar to the work of the railway turnout, IP-MPLS, cisco's CEF adopts similar ideas, but all-optical implementation is faster! The separation of addressing and forwarding extends to the entire network! The addressing network signaling network can be implemented based on Sdn.
In the end, TCP/IP remains the core of the Next Generation Network, TCP/UDP /... complete terminal intelligence, implement IP address logic addressing, configure the underlying transmission network and transmission network based on SDN, such as the entire optical network to complete the actual data transmission, IP addressing is actually equivalent to finding the path to the destination on the map, and how to keep it in mind, or installing navigation on the car. This is done by SDN, and the transmission network, for example, the all-optical network is the transit expressway to reach the destination, and automatically redirects to a specific fork, because addressing has been completed before the High-speed, and the rest is just moving forward according to the addressing result! As for what to do at the destination, it is all about the passengers themselves. It has nothing to do with the highway and the drivers themselves. This is about the transport layer protocol such as TCP. TCP/IP still works in the original way. In the next generation network, TCP/IP is closer to its original form because it removes all the so-called "Complex Intelligence ", remove all the so-called "Tips "...

This is a summary. According to the author, the network is divided into the age of resource shortage and resource filling. In the age of resource shortage, many tips were added to the network outside the core of TCP/IP. In the age of resource filling, TCP/IP came back to its original design, in addition to these tips, it will always be the core of the Internet, but in the Age of resource filling, IP is removed from data transmission as an addressing signaling network. It is worth noting that in the all-optical network era, optical networks no longer exist as a link layer like ATM and X.25, and all networks themselves. The differences are shown in: