All along, as long as we talk about operating server applications, whether it is a web application or a mobile application terminal, we will think that the future is the world of cloud services. Companies such as Amazon, Google, and Microsoft are developing new tools to make server applications easier and more convenient to operate. Therefore, for users, AWS, GCP and Azure seem to be the best choice. After all, these three cloud service tools are convenient and cheap, and can be easily fully automated and flexible. I really want to say that there are still many advantages.
First, it cannot meet the long-term expansion and upgrade needs. Developing a high-quality web application that can be extended and upgraded, especially in the cloud service space, is quite difficult. Even if the application was successfully developed in the early stage, the later expansion and upgrade will require a lot of manpower, material resources and financial resources. To take a step back, even if you succeed, you will eventually reach the limits of cloud technology and the web page itself. The computer's computing speed and storage capacity will grow faster than the network bandwidth. Although for most companies, you don't have to worry about this issue now, but you have to face it. At this stage, the total amount of data in the network is increasing dramatically. After all, our resolution requirements have changed from HD to 4K to 8K. Soon, virtual reality data sets will become popular.
This is a problem, largely due to the way we organize and arrange the network. There are many customers who need content and programs, but there are very few servers that can provide these content and programs. For example, if someone shares a picture of a cat on Slack, and the 20 people sitting around me want to see it, they can only download the picture. In other words, the server has to send 20 times in the same picture.
If the server is now moved to the cloud, which is on a Google or Amazon data center computer, then the network near these areas requires a large throughput to be able to process all the data. At the same time, a large number of hard drives are needed to store everyone's data, as well as a large number of CPUs to deliver data over the network to everyone in need. However, since the rise of streaming services, the situation has been a bit bad.
All of these activities require a lot of energy support, making the entire system inefficient. Not only is it costly, it also harms the environment.
Second, it is too concentrated and vulnerable. Another big problem with centrally storing data and programs is usability and durability. What if Amazon's data center is under traffic attack and is affected by natural disasters such as asteroids or tornadoes? Or, not so exaggerated, what if it happens after a period of power interruption? The data stored on the data center machine is not only unusable for a short period of time, it is even more likely to be permanently lost.
Perhaps someone would say that storing these data separately can solve the problem. But in fact, it won't. Separate storage will only mean more data centers. Yes, separate storage is a risk of reducing accidental losses. But what about the really important data? How to do safe storage? Currently, these data resources are stored in the cloud such as Facebook, Google Drive and iCloud. But what if these companies go out of business and the business stops? What should I do? Even if they never close down, there are limits to using data after storage. Because you have to rely on their services to get this data. Even if you just want to share it with your friends, you still have to pass their services.
Third, it requires us to give us trust, but it does not give us any guarantee. How do your friends believe that what you receive is what you sent? The only way is to believe in the middleman. In most cases, this is okay. However, the websites and networks we use are operated by legal entities registered with the national authorities. The state department can force these legal entities to do anything that is in accordance with their own wishes and in their own interests. Although in most cases, there are some good things that help fight crime or remove illegal content, there are also bad things that abuse power.
Just a few weeks ago, the Spanish government tried its best to use the power in its hands to stop the independent referendum in Catalonia. For example, block all information websites that guide people to vote. Although for most Westerners, freedom of speech may not be the most important issue, it is worthwhile to guarantee the free and open Internet. After all, in essence, the Internet should be free to open, and the content that the user ultimately sees and the content originally published by the author should be verified.
Fourth, it makes us and our data an object that is easily deceived. The biggest drawback of the highly concentrated Internet is the accumulation of personal data. On the one hand, large companies do provide the services people need; on the other hand, they hold the data of all consumers. The data is enough for them to speculate on the future behavior of the consumer, such as what to buy, who to vote for, whether to buy a house, or even want a few children. You know, with this data, it is entirely possible for those who are not willing to use your name to do credit cards, make loans or even buy a house.
Of course, you may think that there is no problem in providing this data. After all, these companies are good enough. You trust them enough to be willing to give your own data. But in fact, the key to the problem is not in them. Yes, they don't have to worry. But what if there are people who are trying to misbehave? Just earlier this year, a very serious data breach occurred in Equifax, a well-known US credit reporting agency. Specifically, a total of 140 million users of data were lost. Now, all of this data is made public. Some people may say that this kind of thing happened once in a decade. If we can be careful and cautious, it can be avoided. But now, the situation is getting clearer and clearer. Data leakage like this, it is very difficult to completely avoid it. But if you let it go directly, it is very dangerous. Therefore, the only feasible solution is to avoid centralized access to user data information from the beginning.
So, what will be the success of the cloud era?A network that is supported by a client server protocol to a large extent and with centralized authorization as a security guarantee not only has various defects, but also problems that are difficult to solve or even impossible to solve. Therefore, we are now looking for a more efficient solution. In the new mode, no one else can store your personal data, and large media files are distributed throughout the network. Moreover, the entire system is completely peer-to-peer and no longer requires a server.
In response to this piece, I have read a lot of articles about emerging technologies. Therefore, I am convinced that peer-to-peer, that is, equal, is the inevitable development trend of the future. Peer-to-peer networking technology exists to replace the existing network components with various protocols and strategies to solve most of the problems listed above. Their goal is to achieve highly distributed data storage. Every customer who is in the network and participates in it will store some of the data in backup form.
If you've heard of bitstreams (that is, BitTorrent, a content distribution protocol), then the following should not be unfamiliar. In this protocol, network users sharing large data files are divided into smaller blocks (each with a dedicated ID) that do not require any form of centralized authorization. To download a document, you only need a "magic" number. The user then finds the people who own the same file and downloads them until they get all the content in the file.
One of the most interesting places is how to find someone with the same file. The bitstream uses a protocol called Kademlia. Under this protocol, each point in the network has a unique ID number that is the same length as the dedicated block ID. This ID can store data information within the block at a certain node with the specific ID closest to the block ID. For random block IDs and network endpoint IDs, the storage distribution across the network should be consistent. These blocks are all content-processable, so it's easier to verify the content.
In addition, an interesting part of using fingerprints to process content is to embed the ID of one of the blocks into another content block, linking the two blocks in a way that cannot be tampered with. If the content of one of the pieces changes, its ID will change accordingly, and the connection between the two will break down. Once the internal connection changes, the ID of the other block will also change.
This ID embedding mechanism helps generate similar blockchains (like the blockchain that supports bitcoin and other virtual currencies) and even generates directed acyclic graphs (ie, Directed Acyclic Graphs, below). More complex structures such as DAG). This graphical connection is named after the inventor Ralph Merkel and is therefore called the Merkel DAG. One of the most typical examples is the git local repository. It uses the giant Merkel DAG to store historical data, all entries and files in chunks.
This brings up an interesting topic, that is, distributed data storage based on content processing is immutable. In other words, the content stored in these blocks will not change. Instead, the modified version of the content is stored next to the existing version. Reusing the contents of the modified version does not cause the blocks to change because their IDs are still the same. This means that in such a storage system, the target file cannot be copied. Go back to the beginning of the article, which means that each cat's picture is unique and will only exist once.
Agreements such as Kademlia, coupled with Merkel blockchain and Merkel DAG, provide us with a variety of tools to form a complete file hierarchy and modify the timeline for large-scale sharing in peer-to-peer networks. At this stage, there are already protocols that use these technologies to build distributed storage systems based on our needs. Among them, the development is better than the number of IPFS.
Problems with names and shared things
So, here, we can already use these techniques to initially solve some of the problems listed above. The entire process almost solved the availability, capacity, durability, and content verification issues. But at the same time, it also needs to solve the bandwidth problem. Since each point can send data to each other, there is no large hotspot or bottleneck.
In the future, we need to be able to expand the upgraded computing resources, which is not difficult. At the moment, everyone's laptops and mobile phones have a variety of features that far exceed the features that most applications require. For example, complex machine learning calculations. Therefore, from a horizontal perspective, the calculations are indeed scalable. As long as we ensure that all equipment meets the necessary needs of users, there should be no major problems.
In other words, in this case, the picture of the cat I want to see is from a colleague sitting next to me, not a Slack server. Of course, if I want to actively share a cat's picture now, the first thing to do is to update the specific channel. That's right, this step sounds very simple. But in fact, it is the most difficult part.
The most difficult part: channel updateSomething will change over time, just to let people have a concept of order and stability. As for this thing, we can think of it as an identity and a name that can show different values ​​over time. But these values ​​are static and immutable. This is a more natural way to regulate the information in your computer and thus has more natural results. If I tell you something, it can't change the fact that I have told you, or force you to deliberately forget what I said. The Slack channel represents an identity. The different values ​​it exhibits over time are gradually increasing information.
But the real problem is that in this channel, we are not isolated. There are too many people who want to share information and change channels. In addition, sometimes these two situations will happen at the same time. Therefore, someone must stand up and decide the final result.
In a more centralized system, there is a single mechanism to determine these update processes and serialize these events. But in a distributed system, everyone is equal. Therefore, there is a need to develop a mechanism to ensure that the entire network agrees on the issue of “world historyâ€.
Today, all the applications we use are supported by distributed networks. The most difficult problem for such a truly distributed network is to reach an agreement. This not only affects parallel updates, but also affects other updates that need to occur. This problem is especially difficult for databases. At the same time, it will also have an impact on other important services such as the DNS. Registering a person's name for a particular block ID or ID sequence in a decentralized manner means that everyone involved in the process needs to agree on the existing name and its meaning. Otherwise, there will be a situation where two different users see two different files with the same name. Processing based on content can solve this problem for the machine, but it cannot solve this problem for real people.
To solve the consensus problem of distributed networks, there are several strategies. The first is to form a smaller team of managers and further select the “team leader†to take the decision. All changes that occur in the process go through the manager. Obviously, this is a centralized system.
The second is a system based on workload proofs like the Bitcoin blockchain. The system solves the problems together through the endpoints, writes the update plan, and has certain guarantees on the consensus issue. Although the problem is difficult to solve, it is easier to check. If there is a disagreement, it can be solved with some additional rules. However, there are several other distributed blockchains that use a consensus-based consensus system. This system can reduce the investment required to solve a certain problem.
Third, it is related to the conflict-free copy data type. However, in some cases of this type, the issue of consensus does not have any impact at all. If all updates are just "add one", then as long as we guarantee that each update is used only once, the order is not important, as the results are the same.
In short, so far, there is no clear answer to this question. After everyone's efforts, there have been many interesting and funny solutions. You only need to choose a relatively balanced solution according to your specific situation.
Privacy issues with public documents in the network environmentObviously, among all the issues listed above, what needs to be addressed is privacy. Since we need to store content in such a peer-to-peer distributed network, how can we not disclose this content? If you can hide it, then content processing storage is certainly a quality solution. After all, if you want to find something, you need the content "fingerprint". Therefore, we need to pay attention to three different levels of privacy issues: open, hidden, and private. As for the third, private privacy, the appropriate solution should be encryption.
Although at first glance, relying on encryption to ensure the privacy of content is indeed a little risky. But in fact, it is not as bad as it is now. Usually, companies and governments use some methods to store sensitive data in a way that is not shared with the public, and to protect it by encryption. The only people who can see or use the data are the confidential personnel hired by the company and the government. Once you have access to the system that stores the data, it often means you can see all the private data.
But if we store private data in a public way, we have to take protective measures to prevent anyone else from seeing or using it. To put it bluntly, this is a truth with some open source security applications.
In the process, if you grant someone the right to get data, he can permanently have the private data after the modification. Therefore, you need to constantly modify the encryption key. The bad thing that can happen today is that as long as someone can get the encrypted data, he will copy it privately.
The end of the cloud era, is the foggy era coming?Despite the many problems that need to be solved, the end of the cloud era is still exciting. First, from a technical perspective, we should be able to optimize peer-to-peer networking technologies. Content-based data storage provides cryptographic authentication for the content itself without the need for reliable authorization. Even in developing countries, big progress can be seen, and they will gradually abandon the data center.
To some extent, the data center will even become a past tense. At this stage, the functions of electronic consumer devices are becoming more and more powerful. Computational and storage functions are almost everywhere.
For companies running web applications, this change should enable companies to help reduce costs, reduce the challenges of developing high-quality digital products, and shift the focus from reducing downtime to increasing customer value. We still need servers in the cloud, but they are just a few of the similar endpoints.
Second, another big benefit to the company and its customers is the processing of customer data. Since there is no need to centrally store a large amount of information, there will be no large amount of data loss. For a long time, software engineering experts believe that the Internet design of customers sending data to company programs is a major flaw. In contrast, companies should send programs to customers to use their programs to process their private data and avoid direct sharing of data. While this model seems safer, it does not prevent companies from collecting the customer data they need.
UV Screen Protector,UV Curing Machine,UV Curing Screen Protector
Shenzhen Jianjiantong Technology Co., Ltd. , https://www.tpuprotector.com