James kretchmar, chief technology officer of the Internet content distribution network developer, Inc., expounds and analyzes the company’s business, the impact of artificial intelligence on Internet performance, and the future development trend of artificial intelligence.

The inspiration for the intelligent edge platform of the company comes from a conversation between Professor Tim Berners Lee of MIT and Dr. Tom Leighton, CEO of the company. James kretchmar, chief technology officer of the company, explains how this is the idea for a decentralized Internet model that has been deployed in more than 1500 networks to improve performance.

Kretchmar said, “to achieve good performance on the network, we may encounter these difficulties in the future, and the original architecture will have server bottlenecks and congestion overload.” So Dr. Leighton and his students began to work on this problem, and they developed algorithms to solve these problems, which is the basis of what we now know as the akamaiplatform.

The essence of these problems is that if an enterprise has a very popular website and end users all over the world try to visit the website, the content must be spread through many different networks to reach end users in other countries, which leads to the problem of slow access.

One thing that may not be obvious about the global Internet is that there is a bottleneck between these networks. If the content of its website is visited heavily, it will burden the hosting server and the surrounding network. This is the reason for the driver model, which is a highly distributed platform. In this model, if end users request a piece of content, they will be sent to the server closer to them. This avoids these network bottlenecks, improves performance, and means that end users do not have to send all of them to a centralized server. “

The influence of artificial intelligence and machine learning

Kretchmar described how artificial intelligence (AI) and machine learning (ML) affect the performance of the global Internet, and how AI can help network management with the continuous development of the global Internet,

“Part of the way we make services work is to understand the performance of the global Internet in real time and to deliver content to users by taking advantage of the best and changing paths on the global Internet,” he said. Machine learning and similar technologies can help us understand the structure of the global Internet and the different characteristics of different paths in history and now.

Now there is another challenge, that is, there are too many robots trying to carry out malicious attacks on websites. They want to steal data and try to destroy login credentials, in which case machine learning is very important to figure out which are robots and which are humans.

However, nowadays learning strategies have become more and more complex, so that these strategies actually do not have to try to identify whether it is a machine, but to determine whether it is a real human. We can see what the interaction from real human is like, which can explain many different factors. For example, if it’s a mobile device and they input data through their mobile phone, it will move the location, and it can also conduct similar checks on desktop devices. “

5g and Internet of things

Looking ahead to the next decade, the chief technology officer of the company expects that AI will be used to better utilize the emerging 5g technology and data of the Internet of things.

“There will be more and more Internet related data, such as 5g, that will help to achieve the Internet of things, and the challenge will be to use the growing data to do something,” he said. For example, our work on intelligent routing system is to learn data from basic methods, and it becomes more and more complex. But when we observe what is going to happen, more websites use high-definition video, and more end users need better network connectivity. This will make the challenge more complex, so we will need more intelligent algorithms to process the data to find out how to provide the best performance. “

Guard against threats

Another AI related trend predicted by kretchmar involves protecting websites from evolving cyber attacks.

“Today, the ecosystem of the Internet is becoming more and more complex, including evolving cyber attacks,” he explained. Malicious attackers have been looking for a more creative and intelligent way to attack in order to try to use the system or service provided online, so machine learning technology is involved in it.

We recently launched a product for the way modern websites work, that is, they usually absorb a lot of third-party content to make their websites work normally. Therefore, if you visit large tourism companies or retail websites, their websites are built by extracting a lot of content from different places to make their main websites work normally.

What network attackers do is become creative, they will damage some content of the third party, and the third party will include part of the data in the delivery of the main site. Here, network attackers may implant code, such as trying to steal credit card information, while the owner of the main site does not really realize that the third party is threatened, and also contains malicious code in its site.

Now, what we have developed is a way to fight against it. By understanding the request flow and automatically detecting whether the malicious behavior exists, we can determine whether the website is having malicious behavior. In this way, website owners can receive alerts and take appropriate actions. “

Editor in charge: CC

Leave a Reply

Your email address will not be published. Required fields are marked *