A brief talk on game operation--game Cloud series Three

A game product on-line, only from the technical angle, divides into the software level game code development, and the hardware level code deployment on-line. The development of split code does not say that the deployment of the game on the line, become a problem we have a headache. Why a headache? Some reports show that the majority of the game life cycle is only 3 months. In accordance with the formal on-line process, from buying servers, loading environment, into the IDC room so down, just put the game online, but the game is not to force. This toss down, how much of our youth waste? How much of our hard-earned money has been wasted?

There are several ways to deploy a general game:

1, managed IDC room deployment.

2, agent deployment.

3. Leasing VPS Environment deployment.

4, leasing cloud host environment deployment.

For the first, IDC room deployment, I think is the most wasteful. If the purchase of their own servers (can also be rented), the company's procurement approval process, server transport over, install system environment test, IDC server shelves, etc., this is a very cumbersome process.  Especially in the later operation of the maintenance of the time, when the server is abnormal, want to restart the server, this simple requirements become a luxury. For the second, the code to the agent deployment, I think is the most unsafe. You do not have to care about what kind of server you buy, how to configure the environment, etc., you only need to put the game code to the agent, the later deployment of the online, operation and maintenance management are not you worry. Seemingly quite worry, but the database and so on, are others to help you operation and maintenance management.  Your code, data are given to others, how to talk about security? For the third, leasing VPS environment deployment, I think is the performance and availability of the worst. VPS, everyone is very familiar with. is to virtualize many systems on a single server, and then split it up with multiple users. Although theoretically, each VPS can be allocated independent public network IP address, independent operating system, the implementation of different VPS disk space, memory, CPU resources, processes and system configuration isolation. For example, common virtualization technologies are: VMware, Xen, KVM, etc. But the reality is that when a certain VPS resources are used too high, this will affect the performance of other VPs.  Eventually lead to multiple VPS Rob resource phenomenon, such as allocating a 4 nuclear 4G VPS Space, in the end, you may not even 2 nuclear 2G resources can not be robbed. For the fourth, many people think that there is no difference between a VPS and a cloud host. In fact, there is a big difference, about this piece of contrast, in this do not do too much introduction, have free words, and then talk about this piece of content. The biggest difference, now can be simply understood in the resource isolation this piece, is the cloud host and the cloud host resources use, does not dare say hundred cent isolation, but at least at the bottom of the isolation of this piece, do is still relatively good.

So the cloud host compared to several other deployment methods, in the operation of the block has the following advantages: 1, cost considerations

Considering cloud services, cost factors are the first factor to consider. If who said, money is not a problem, then there is no problem, hehe. Compared to the traditional IDC, cloud services can effectively save on the physical hardware input costs, as well as in the operational dimensions of human costs. Why can we save manpower costs in operation and maintenance? For example, we have deployed the application to the cloud, we do not care about the rear cloud server monitoring, security and operational management, all this by the cloud to help us fix.

And for many games on the line, the early days do not know how many users will be after the online. If you blindly put the cost into this piece, it is likely to cause losses. Many company's game on-line, is chooses the cloud service to deploy the game on-line, carries on the prophase "the mapping". If the user is large at the back, simply increase the cloud host. If the user is low, we reduce the resources deployed.  This can effectively control costs, flexibility is very high. 2. Network stability

In order to ensure the fluency of the game, strong network protection is essential. Compared to the traditional dual-line room, cloud services use BGP network, to our network to add a strong guarantee. Become a solid and strong backing of our network. 3. Rapid Deployment

Rapid deployment, performance in several ways. First, at the hardware level. We do not need to buy a server, do not install the system, the cloud host to use, rapid deployment.

The second rapid deployment, at the system level. Generally in the cloud service level, there are mirroring and API functions. The combination of the two, even if the creation of 1000, 10,000 servers, is an instant thing. What is mirroring function? For example, I now have a new area online, I put the environment are installed in the cloud host. I've packaged the system environment as a mirror image (a bit like an ISO image), and then when I create a cloud host, I can use this image to create a system with exactly the same environment and data. Through our open a cloud host, we generally are in the Web console click operation. Of course, we can also use the API interface, the code to create a cloud host, the real meaning of the rapid deployment. 4, disaster preparedness, data backup

The general cloud host uses simultaneously writes many data to guarantee the data security. The system environment of our cloud host may exist on a physical machine, but at the same time the data of the environment is redundant on other physical machines. When the physical machine of this cloud host is down, it will automatically migrate to the redundant data on other physical machines to ensure the reliability of the server.

5, Operation and maintenance monitoring

Traditional operation and maintenance monitoring mode, in the application layer monitoring, such as the use of monitoring Polaris monitoring a URL to monitor the availability of a page service. At the server level, usually install some monitoring software, such as cacti, Nagios, Zabbix, and then build their own operational monitoring Center, monitor the availability of the server and applications. and cloud services, all with these monitoring functions, we only need to fool in the Web page settings to monitor the properties, eliminating the complex to build a monitoring environment, convenient and efficient.

6. Anti-attack

In the prevention of attack, but also to give you surprise constantly. The bottom layer uses a hardware firewall to prevent attacks, in the Web console, we can also set some of the attack thresholds and parameters.

Original Author: Architecture Cloud Shoke article from: http://game.aliyun.com forwarding please indicate the source