Week6
Technology: Transport Control Protocol(TCP)
Welcome to Week 6 of IHTS. We are in our second week of the more technical bits of the course. I try to keep the technical bits interesting and not go into too much detail.
This week, we’ll be covering transport control protocol (TCP). The Transport layer is built on the Internetwork layer and is what makes our network connections reliable.
hello, last week we learned the first major Link layer and the second layer Internet Protocol (IP), then let us continue to learn the third layer Transport Control Protocol.
Transport Layer
By prior study we already know that the main role of the network layer: defining a network address, to distinguish between network segments, MAC addressing subnet, on different subnets for routing data packets. It also has a magic, there have been no long-term storage warehouse in the internal network, all long-term storage are outside of the network, which is the IP transmission speeds so fast reasons. But because it only takes into account on the one hand, it is also flawed in other ways. One of these is possible in data transmission, there are cases of data loss, data loss case if there is, we need a layer to be responsible for the recovery, and this layer is what we want to speakers today Transport Layer.
TCP layer is based on the IP layer, one of its aims is to compensate for errors that may appear in the IP layer, and maximize the use of available resources. TCP / IP networking to efficiently transport, so we need to know our network is fast or slow, this is what we use TCP layer to solve the problem. How fast the underlying network, how high reliability, if something goes wrong, how we should deal with. Thus, the core idea of TCP / IP is the protocol when transmitting data, dividing the data into a plurality of packets, and each packet will be sent out. Then we keep them until they are confirmed on the other side of the replies, and then we only throw them away. In some cases, if the packet loss, it may be sent again, until the final object in the system so far.
In practice while such lost data packet recovery after confirming that the network is not much storage of requirements. Yet the storage requirements on the design of the terminal. When we put computers, laptops, mobile phones and other connected to the network, they themselves are responsible for saving copies of data, they do not expect the network to help store work. This is an excellent mechanism is one of the reasons for the success of the Internet .
As more and more connected devices, the network becomes more and more congested, the network backbone speed becomes slower and slower, and so began to fail. So in the 1980's, there is a prophecy says the Internet is about to fail. And this view is supported by many people in the past, because they believe that academia can not afford to make the Internet continue to develop. At that time people even think let IBM and other electronics manufacturers to engage in the network, other people will spend money provider's network. As the saying goes the times create heroes, in this desperate backdrop, engineer Van Jacobson points out, he created a patch to solve network congestion problems. Save the Internet, a really amazing genius exquisite ah. Then let us introduce introduce him.
Van Jacobson - Slow Start Algorithm
Rotary Building is going to dump, turn the tide that is down, that's can be described as Van Jacobson, saying the crisis intensified at the time of the network, its slow start algorithm and TCP flow control algorithm to save the entire online world. NSFNet and Ethernet for university network is very popular at the time, many people through the exchange network, send and receive e-mail, people are thrilled to this new technology. But more and more computer access to allow increased network load, slow speed, a lot of packet is lost in transmission.
Slow start , congestion control mechanism is a transmission control protocol used. Slow start also called exponential growth. Slow start means that will increase each time the TCP receive window to acknowledge receipt. Increased size is the number of segments have been confirmed. This situation is maintained until either did not receive some segment or window size reaches the threshold defined in advance. If a loss event occurs, TCP will think this is network congestion, it will take steps to mitigate network congestion.
Van Jacobson was a researcher at Lawrence Berkeley Laboratory, it was the mid-1980s, each course has a news group, like a small news group, all jobs will be put online. Van try to learn from LBL's office to download course material on a machine Berkeley Evans Hall, he found that network throughput is zero, this is undoubtedly too bad. He found Mike Karels, found Mike Karels also deeply troubled by this issue, the two decided to work together to solve this annoying.
Network Throughput : data network is composed of a data packet, the firewall process for each data packet to be resource-intensive. Is the maximum throughput rate without frame loss, the device can accept
The easiest time running TCP / IP approach is to start Berkeley Unix, because it is embedded in a very good application ARPA-funded, but the performance of this program is very poor, small-scale test of time during the collapse, Van worked for them months to find where the problem, but there is no breakthrough, this time Van think this is due to ignorance of the line is how it works, if we can understand how it works, and that naturally has the appropriate treatment method, so they the focus shifted to configure itself and protocol aspects of the agreement.
TCP congestion control
In time, Van raised such a mode, as shown in FIG. Lateral represents the time, the network data transmission will take some time. Different longitudinal containers of different thickness on behalf of bandwidth in the picture is ten times the gap, actually up to several hundred times. Relatively thick, on behalf of the network at both ends, one end of a transmission source, and the other end is a source object. The sender needs to send a large packet, if once let them into the network may cause congestion. In this case the above-mentioned Acknowledgment (hereinafter, simply referred to as the ACK ) is very useful. We control proceeds speed packet network, you receive a source object whenever the ACK, the transmitting side we retransmission packet. ACK fact, like a clock, like, suggesting that when the network is smooth, then you can send a packet without causing network congestion. After this mechanism up and running, the new packet is sent will not let bottleneck (network) packet in causing congestion.
This is a very good model, good running in one week after the exchange of packets, but this is the emergence of a problem. How to start it? This requires the above-mentioned slow start algorithm . If you are using a slow start, gradually start this mechanism, it slowly run up, then we can prevent network congestion, which is the target of TCP. About the slow start algorithm, the video is not without its elaborate too much, so I'm looking for some information, you can try to refer to the reference.
- TCP slow start algorithm
-
TCP congestion control algorithm slow start and congestion avoidance algorithm
而如何让这种机制植入到全球的TCP/IP中的呢？Van使用了一个比较偏门的办法，他让组内的资深内核黑客编写了一个程序，这个程序会获得用户数据报，并发生内核错误，他们把所有这些程序打包 放在TCP/IP的邮件列表上，很多人会去下载试用，于是程序崩溃，Van就获取他们的反馈信息，并将其修复，再崩溃，又修复，如此循环，最终生成了一个不会发生内核错误的版本。
The Domain Name System
Domain Name System，域名系统，是一个为任何连接到网络的电脑、服务器、或其他任何资源命名的分层分散式系统。它将各种信息与域名分配给每个参与的实体。最突出的特点是它使用对人类友好的、更容易记住的域名去代替数字的IP地址，实现定位和识别计算机服务和设备与底层网络协议。通过提供一个全球分布式目录服务，域名系统是因特网功能的一个重要的组成部分，并且自1980年代以来一直在使用
为什么会存在域名系统呐？虽然因特网上的节点都可以用IP地址惟一标识，并且可以通过IP地址被访问，但即使是将32位的二进制IP地址写成4个0～255的十位数形式，也依然太长、太难记。因此，人们发明了域名(Domian Name)，域名可将一个IP地址关联到一组有意义的字符上去。用户访问一个网站的时候，既可以输入该网站的IP地址，也可以输入其域名，对访问而言，两者是等价的。例如：微软公司的Web服务器的IP地址是207.46.230.229，其对应的域名是www.microsoft.com，不管用户在浏览器中输入的是207.46.230.229还是www.microsoft.com，都可以访问其Web网站。
域名的分层
除了代表各个国家顶级域名之外，ICANN最初还定义了7个顶级类别域名，它们分别是.com、.top、.edu、.gov、.mil、.net、.org。.com、.top用于企业，.edu用于教育机构，.gov用于政府机构，.mil用于军事部门，.net用于互联网络及信息中心等，.org用于非赢利性组织
随着因特网的发展，ICANN又增加了两大类共7个顶级类别域名，分别是.aero、.biz、coop、.info、.museum、.name、.pro。其中，.aero、.coop、.museum是3个面向特定行业或群体的顶级域名：.aero代表航空运输业，.coop代表协作组织，.museum代表博物馆；.biz、.info、.name、.pro是4个面向通用的顶级域名：.biz表示商务，.name表示个人，.pro表示会计师、律师、医师等，.info则没有特定指向
TCP Wrap Up
，
总的来说，TCP做的事就是弥补IP层所做的工作的不完美之处。数据到达时，可以不再是无序的；TCP使得数据在发送端储存了拷贝，如果没有收到ACK就会重复发送，确保数据不会丢失；TCP的缓存机制，使得互联网快速发展。
PS：（这两周都有些写课后思考，但是由于上一周内容太多，忘记加上去了，所以就决定把这两周的课后思考发在技术层次的最后一周，也就是下一周。）