All Products
Search
Document Center

Cross-region load balancing

Last Updated: Sep 11, 2019

Multiple IP addresses of an enterprise application service may be distributed in data centers of different carriers or manufacturers in different regions. In this case, a single IP address cannot bear all users’ access requests and it is very hard to build a load balancing architecture for the application service. There is no other service than DNS that can simply and efficiently organize IP addresses of multiple data centers to provide service for customers.

However, DNS itself cannot sense availability of IP addresses. Thus, it cannot route the access traffic of a user to the available IP address of the application service quickly and efficiently in case of faults or disasters.

Here, we provide a cross-region multi-IP-address load balancing solution based on GTM. These scenarios can help you design the service architecture.

Scenario 1: cross-region even traffic allocation

The enterprise application service has multiple IP addresses. When the service capability, computing capability, and network bandwidth of all the IP addresses are the same, the even traffic allocation principle can be applied to evenly allocate the traffic for these IP addresses. Hence these IP addresses simultaneously bear the access requests when users access the domain name of the same application service. The principle is illustrated below:

Schematic diagram

The implementation procedure is described below:

Step 1: make global settings

Make global settings for a GTM instance according to the figure below:

全局配置

Step 2: configure load balancing for the addresses

In a GTM instance, configure load balancing for the address pool according to the figure below:

地址池管理

Step 3: configure health check

In a GTM instance, configure the HTTP health check for the address pool according to the figure below:

HTTP健康检查

Step 4: configure a global access policy

In a GTM instance, configure a global access policy applicable for all users according to the figure below:

访问策略

Step 5: configure CNAME access

In the domain name resolution configuration area, complete the CNAME access configuration for the application service according to the figure below, so that the application service can access the GTM instance.

CNAME域名接入

Scenario 2: cross-region traffic weighted allocation

The enterprise application service has multiple IP addresses. When the service capability, computing capability, and network bandwidth of each IP address are different, the traffic weighted allocation principle can be applied. You can set different weights for different IP addresses, so that the IP addresses bear the users’ access requests at a certain ratio when users access the domain name of the same application service.The principle is illustrated below:

Schematic diagram 2

The implementation procedure is described below:

Step 1: make global settings

Make global settings for a GTM instance according to the figure below. NOTE: Select weighted round robin when configuring the balancing policy.

加权轮询

Step 2: configure load balancing for the addresses

In a GTM instance, configure a weight for each IP address in the address pool according to the figure below:

地址池-权重配置

Step 3: configure health check

In a GTM instance, configure the HTTP health check for the address pool according to the figure below:

HTTP健康检查

Step 4: configure a global access policy

In a GTM instance, configure a global access policy applicable for all users according to the figure below:

访问策略

Step 5: configure CNAME access

In the domain name resolution configuration area, complete the CNAME access configuration for the application service according to the figure below, so that the application service can access the GTM instance.

CNAME接入