Experiment Scene:
Xian Lingyun System High-tech Co., Ltd. using IIS to build a Web site, the domain name for nlb.angeldevil.com. As the business gradually increased, the website speed is also more and more slow, and often failure, for the interests of the company brought a lot of inconvenience; the company decided to use two Web sites to provide access to clients. So we use the Network Load Balancing technology. The IP addresses for both servers are: 192.168.1.10 and 192.168.1.20, and the cluster uses an IP address of 192.168.1.30. To manage the cluster on a host of two NLB clusters: Add a network adapter to each of the DC and DC (two server) members, respectively: 10.1.1.1 and 10.1.1.2.
Objective: To understand the concept of Network Load Balancing: Completing the configuration of NLB servers: Verifying that NLB servers are successful: Best practices for Network Load Balancing:
Experiment topology:
Experiment steps:
The concept of Network Load Balancing:
Two types of server clusters can be supported in Windows 2003: Network Load Balancing (NLB) and server clusters (CLUADMIN);
In this chapter we explain Network Load balancing:
Benefits of Network Load Balancing: Network Load Balancing can enhance web, scalability and scalability of some services, such as ftp: We know that the reliability and scalability of the application services provided on a single computer are limited, but we can do this by running a service-load-balancing technology with two servers or more computers! Network Load Balancing clusters can be used in Windows ' domains or in the work done under Windows: When our server because we think that the failure of the time we can use one of the other to complete the network's normal work; but we do so by the access side of course is not feel! Here we use the Web site as a demonstration of Network Load Balancing deployment and build: Although we can work with a lot of servers when we're done with load balancing, but we need to know that Network Load Balancing can support up to 32 computers:
Second, the basic configuration of the NLB server:
To facilitate the integrity of our experiments, the environment we belong to is the domain environment under windows; We've installed the IIS service before we do the experiment. ; we have created Nlb.angeldevil.com records on the DNS server; Of course we can add them in the Hosts file, but our best approach in DC is to add resource records in DNS:
There are three main steps in configuring Load Balancing: (1), enabling Network Load Balancing.
(2), connecting to an existing cluster:
(3), adding a host to the cluster:
1.1, enable Network Load Balancing: In the beginning---run the input nlbmgr, click the "OK" button, open the Network Load Balancing Manager window;
As shown in the figure:
1.2. Right-click Network Load Balancing cluster, click New Cluster, and then in the dialog box that pops up the IP address and other cluster information, select cluster operation mode as multicast, and then click Next. (Note: The IP address we are adding here is the same network segment as the DC, and the reason we use a multicast here is that the client is receiving information at the same time)
1.3, in the "Cluster IP Address" dialog box we can add IP address ; If you have more than one cluster IP address, we can also add multiple IP addresses: as shown in the image is the IP address of the added port: (note: In this trial we have only one IP address for a network cluster so we choose to add the following automatically instead of all ports. And we are parsing the time we are through DNS server resolution and not using other protocols so we use the TCP protocol port HTTP80; so choose 80 here ok!!! )
1.4, after the completion of the above cluster IP address planning, will appear as shown in the port rules: that is the port rules? A port rule is a communication that can be dispatched to clients according to the amount of load per member in the cluster. Of course, we can also delete the port rules in the image shown below, so the removal of the port rule has the above effect? Removing the port rule also makes it more obvious that the effect of the cluster can be seen by removing the port rule in response to the client request in accordance with the priority of the port, assuming that the primary opportunity for a high priority failure is the service. Of course, for the client, it doesn't feel like the!!! is going to fail. In this experiment we did not delete the port rules but follow the port rules to experiment.