DNS configuration to support an accelerator

About configuring IP addresses in DNS

When you set up Web or streaming server acceleration, you configure the accelerator's IP addresses and IP aliases in DNS so that the appliance can receive requests for a Web or streaming server. However, whether you want the accelerator to handle all requests for the Web or streaming server determines whether you configure DNS to include the IP address of the Web or streaming server that the appliance is accelerating.

Note
For streaming MMS servers, you can map only one host name to one IP address; mapping of multiple host names to multiple IP addresses is not supported.

About configuring the server IP addresses in DNS

The following table shows how to configure DNS, depending on whether you want the accelerator to handle all requests.

If you want...	Configure this in DNS...	Other requirements
The accelerator to handle all requests for the Web or streaming servers	Only the IP addresses of the accelerator, not those of the Web or streaming servers that the appliance is accelerating	The host name for the accelerator and the Web or streaming server must be different.
To split the requests between the accelerator and the Web and streaming servers	The accelerator IP addresses and the Web or streaming server IP addresses	None

Many organizations have the accelerator handle all the requests for the Web or streaming server. However, you might, for example, want to split requests between the accelerator and a Web or streaming server in either of the following circumstances:

The link between the client and the Web or streaming server cannot handle the request load to the Web or streaming server.
Neither the Web or streaming server nor the accelerator can handle the full request load.

Example of the accelerator handling all requests

The following illustration shows a deployment in which the accelerator handles all requests. To support this deployment, you configure DNS with the accelerator IP address but not the Web or streaming server IP address. Notice that in this type of deployment, the accelerator and the Web or streaming server have a different host name.

The flow of the request in the previous illustration is explained in the following table.

Stage	Description
1	For Web acceleration, the client sends the URL http://www.A.com, which the user entered into the browser. For streaming acceleration, the client's media viewer is directed to send a streaming server request to www.A.com.
2	The DNS server resolves host name www.A.com as IP address A.
3	The client sends the request to IP address A, the accelerator.
4	If the accelerator does not have the requested content in its cache (a cache miss), it fetches the content from the Web or streaming server.
5	The accelerator returns the requested content to the client; it returns the content immediately if it has the requested content in its cache.

Example of requests split between the accelerator and the server

The following illustration shows a deployment in which requests are split between the accelerator and the Web or streaming server. To support this deployment, you configure DNS with both the accelerator IP address and the Web or streaming server IP address. Notice that in this deployment, the accelerator and the Web or streaming server have the same host name.

The flow of the request in the previous illustration is explained in the following table.

Stage	Description
1	For Web acceleration, the client sends the URL http://www.A.com, which the user entered into the browser. For streaming server acceleration, the client's media viewer is directed to send a streaming request to www.A.com.
2	The DNS server resolves host name www.A.com as IP address A.
3	The client sends the request to IP address A, the accelerator.
4	If the accelerator does not have the requested content in its cache (a cache miss), it fetches the content from the Web or streaming server.
5	The accelerator returns the requested content to the client; it returns the content immediately if it has the requested content in its cache.