Scenario: Using multiple accelerators for a single Web or streaming server

About this scenario

In this scenario, the Web or streaming site is very big and very busy. One Web server cannot handle the amount of traffic. Instead of deploying multiple Web or streaming servers, multiple accelerators were deployed to reduce the load on the Web or streaming server.

Illustration of the scenario topology

The following illustration shows a single Web server or streaming server supported by multiple accelerators.

Using an SLB for request distribution

In this scenario, a Server Load Balancer (SLB) is deployed in front of the accelerators to provide a way to distribute incoming requests over the accelerators.

The SLB takes the name of the Web or streaming site (in this case, www.abc.com) instead of the accelerators taking the name of the Web or streaming site, as in the previous scenarios. When you use an SLB, you do not need to enable transparent proxying. The reason is that the host name in the URL that users enter to reach the Web server site is actually resolved as the IP address of the SLB.

Note
You could also use transparent proxying for distributing requests over the accelerators by deploying an L4 or L7 switch or a WCCP router in front of the accelerators.

Host name and IP address of the three accelerators

The following table shows the configuration of the host name and IP address for each of the three accelerators.

Data to provide	Configuration data for each accelerator	Comments
First accelerator's host name and IP address	www-accel1-abc.com IP address 1	You can assign any host name that you want to each accelerator. For each accelerator, enter its unique IP address.
Second accelerator's host name and IP address	www-accel2-abc.com IP address 2
Third accelerator's host name and IP address	www-accel3-abc.com IP address 3

After setup, you can change the Primary IP Address option (on the Setup tab > Network > Interfaces page) and Host Name option (on the Setup tab > System > Host Name page) in the Appliance Manager.

Rules on each accelerator

The following acceleration rules are configured on the three accelerators.

Acceleration rule on www-accel1-abc.com

Accelerator				Web server
IP	Port	Host	Path	IP/DNS Host	Port	Host	Path
*	80	www-accel1-abc.com	*	www.abc.com	*	www.abc.com	*

Acceleration rule on www-accel2-abc.com

Accelerator				Web server
IP	Port	Host	Path	IP/DNS Host	Port	Host	Path
*	80	www-accel2-abc.com	*	www.abc.com	*	www.abc.com	*

Acceleration rule on www-accel3-abc.com

Accelerator				Web server
IP	Port	Host	Path	IP/DNS Host	Port	Host	Path
*	80	www-accel3-abc.com	*	www.abc.com	*	www.abc.com	*

Configuring the SLB

You must configure the SLB so that it can distribute the requests among the accelerators. How you configure an SLB to know about the accelerators varies according to manufacturer.

The following table shows the data that you need to configure on the SLB in this scenario.

Data to provide	Configuration data	Comments
Host name	www.abc.com	You must assign the SLB the same host name as the Web server. Assigning the same host name to both devices will not cause a problem because they have different IP addresses. You make this host name known as the host name of the Web server.
IP address	IP address P	This is the virtual IP address of the SLB, that is, the IP address that you make known to the world as the IP address for the Web server. However, it is actually the IP address of the SLB.
How to distribute requests	Distribute based on load	The best approach is to configure the SLB to distribute requests based on load. It is likely that copies of the same object will be cached in multiple accelerators. However, this duplication is not a problem when deploying accelerators because the reason for deploying accelerators is to move data closer to users and to serve it quickly. In this scenario, the Web server's data is served quickly from three accelerators.