Shulou(Shulou.com)06/01 Report--

This article is a bit extreme, but it may also cater to some individuals or organizations. This article is short, and I will continue to write on this topic later. The heart hurts too much, the person is too bad.

1. Protocol stack embedded chip

If you think that implanting a chip into the protocol stack is consuming materials and eroding costs, you are wrong. The TCP/IP/Ethernet stack has now become the de facto standard and has not changed for many years. Solidified things in view of circuit design, mold opening, silicon wafer technology and other costs, generally the more constant things to consider curing, otherwise you will have to pay a high price, and is not conducive to protect the customer's investment. In fact, the protocol stack has long been stable and has not changed much, but why is it worth solidifying now? Because the CPU of PC can bear the complexity of software in the past, now it is all micro-terminals, and its ARM CPU can not bear the huge cost of software computing, so hardware computing will be even better. I say this from the consumer's point of view. Although at the beginning, it may be more expensive, but the hardware is like this. After recovering the early investment cost, the subsequent tariff will continue to fall. As for the chip fever problem, this kind of technical problem can definitely be solved. What is difficult to solve is the consumer's inertia problem.

Why many people or organizations do not want to do this, the first is to consider the cost. I don't know much about economics, but I don't know it at all. I know a little bit. What I understand is that at present, big brands are not in the hands of consumers at all, which is a very bad sign. In fact, it is technology that leads consumers. If there is no 1nm technology (exaggerated, people in the department should not be serious), then consumers should not use it. It is as simple as that. It does not mean that consumers need it. Intel and TSMC will certainly be able to do it. Big brands are also not in the hands of Huawei Hayes. Designing a chip is one aspect, and doing it is another. This is not the melee era from the end of the Middle Ages to the 19th century, when you have to make it for me what I need. If you can't do it, I'll find someone else! That era led to the Sun King Louis XIV and Napoleon.

What is worth solidifying is a universal thing. With the popularity of home terminals and the abundance of Internet services, household exports will encounter bottlenecks sooner or later. This bottleneck is not on hardware, but can be broken through with hardware but supported by software. Just wait.

two。 About IPSec

IPSec is going backwards in history. As I mentioned earlier, what a stupid idea to encrypt video streams with IP layer × ×, the network layering model is dead. I am speaking on the premise of my long-term research in this field, and it has nothing to do with any company or individual who is currently paying attention to this business area. If you want to solve a specific problem or be coerced, do it. Technology is metaphysical in the end, and it would be wrong to just follow the script. Like me, I can say and make any details in a field, such as A, but I still say that An is shit, behind which half is metaphysical understanding, the other half is piety.

The topic of security is inherently application-related, not just data security as some traditions think, because the security operation itself will affect the behavior of the data. It is foolish to encrypt a video stream with ECB. Encryption should be done immediately after video encoding, not on an unwitting but sensitive intermediate BOX. I always see that you are trying. The video is just one example, all kinds of examples, full outbreak. How stupid it is to use historical IPSec to encrypt the IP layer uniformly! The IP layer is a simple, simple connectionless best-effort service. IPSec adds two unidirectional SA, which makes an IP stream a two-way stream in the sense of security. How sad. It is true that IPSec was successful in the age of file transfer, but now it is doomed to failure. Security is application-aware and is automatically resolved by the application. There is no unified IP layer security solution, if you want to do it, you need to reluctantly guess the contents of the packet, deep parsing, how tiring.

Why cater to TCP, just because it has so many scenes?

Enter August 2, continue.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.