From http://www.cnblogs.com/allen0118/p/4294066.html
In the large-scale Web application system, due to the large amount of data requested and the factors of concurrency, the web system will be the phenomenon of downtime, the method of solving this kind of problem I personally think mainly in the following areas:
1.IIS load Balancing.
2. Database load Balancing.
3. System architecture optimization, such as Report server and application server separation.
This article mainly introduces the following IIS load Balancing implementation method, the author is also slowly groping, if there are improper places also please the great God guidance below, in order to common progress!!
Demo Environment Introduction:
Server 1:18.13 (the IIS server used to divert).
Server 1:18.49 (the IIS server used to divert).
Server 1:50.32 (servers accessed by the user).
Web site for demonstration: A website called webtest, which is a picture that is sufficient to demonstrate the effect.
Install server Farms as shown in:
The entire installation step is very simple, followed by the prompt to go, after the installation is completed in IIS can see the server farms project, as shown in:
Now we create the server through the server Farms, as shown in:
How many IIS servers are created, and I have created 2 here, which can be tested in the "Run state test" after the creation is complete, as follows:
The server farms determines whether the target IIS server is normal and is judged by the data returned by a file inside the target server, as shown in the following configuration. Health.txt is used as a validation of a file, the contents of which is OK, then if the file returned by the data is Ok,server farms will be judged that the server is a normal state, and vice versa:
For how to balance the pressure of the server, the server farms has provided some algorithms, as shown here, do not do a detailed introduction, we are interested in the words can be tested,
Two IIS server verification success, stating that our configuration is correct, the next step we will test: I directly access the 50.32 server, this time the rendered page is 18.49 of this server image.
OK, now I'm going to stop IIS 18.49 on this server, as shown in:
When 18.49 of this server's IIS stops, we look again at the server state inside the server farms as shown in:
When 18.49 was hung up, we visited the 50.32 server again, and the results came out:
Conclusion: When more than one IIS server is configured, according to our defined equalization rules and algorithms, it will automatically coordinate and allocate the current request to achieve the purpose of the diversion, in the above demo, when 18.49 is inaccessible, the automatic shell switches to 18.13 servers.
PS: Although it is different server, rendering different content, here I am to facilitate viewing the effect, so the use of different pictures to display, otherwise it is not easy to distinguish.
Load balancing under IIS for Windows