Library Table of Contents Index Previous Next Comments

About a GRM L7 server--Application-level (L7) redirection

How a GRM L7 server performs redirection  Top

Like the GRM DNS server, the GRM L7 server includes intelligence to support proximity measurements, detection of unavailable GRM agent caches, and filtering. However, the GRM L7 server can provide more precise redirection than the GRM DNS server. The reason is that a GRM L7 server uses Application-level redirection (L7 redirection). L7 redirection is based on the URL and not on the header content. The GRM L7 server rewrites the request based on the client's IP address, and sends a 302 location changed response to the client. The response includes the IP address of the closest eligible GRM agent cache or caches.

In contrast, the GRM DNS server returns a list of IP addresses of eligible GRM agent caches, and the client browser selects the cache to send the request to.

Note
The method of rewriting varies for HTTP and streaming requests. See GRM L7 server request rewriting for HTTP and streaming content for details.

When you need to include a GRM L7 server  Top

The following table provides general guidelines for determining when you need to include a GRM L7 server in your deployment.

If...

Then...

The local DNS nameserver and the clients are physically close.

You can use all GRM DNS servers (a pure GRM DNS deployment).

If the local DNS nameserver and the clients it services are not physically close.

You will want to include one or more GRM L7 servers in your deployment.

You must provide a means for client requests to be directed to the GRM L7 servers. See Requirement to direct requests to the GRM L7 server.

For example, some enterprise CDNs are spread over continents, with clients located on one continent and their local DNS nameserver located on a different continent. If only GRM DNS servers are used in such cases, the GRM agent cache IP addresses that GRM DNS servers return would be on the same continent as the local DNS nameserver. The reason is that the GRM DNS server sees only the IP address of the local DNS nameserver, not the IP address of the client.

By deploying a GRM L7 server for redirection service to clients, the GRM L7 server will redirect requests based on client IP address. Therefore, the GRM agent cache IP address returned to the client will be truly close to the requesting client.

Requirement to direct requests to the GRM L7 server  Top

When you deploy a GRM L7 server, your deployment must include a way for client requests to locate the GRM L7 server. You can use one of the following methods:

You choose the solution that best meets your requirements. For example, if your network infrastructure already includes a locator device, you can configure that device to send client requests to the GRM L7 server closest to the client.

Section B, Types of deployments with a GRM L7 server describes the advantages and disadvantages of each method and provides deployment examples.

Note
If a proxy.pac file is used to direct client requests to a GRM L7 server, the GRM L7 server might return the IP address of more than one GRM agent cache close to the client. If another method is used, the GRM L7 server rewrites the client request with the IP address of only one GRM agent cache.

GRM L7 server request rewriting for HTTP and streaming content  Top

The L7 request rewriting process and how you can influence it differ for HTTP and streaming protocols.

Rewriting of requests for HTTP content

For HTTP, a GRM L7 server sends a 302 response to redirect a client request and rewrites the URL as

http://redirected-agent-ip/original_url

Normally, a GRM L7 server returns the closest GRM agent cache in its 302 response to the client. However, for content that does not require a large amount of bandwidth (for example, HTML files and GIF files), it is probably more efficient for the GRM L7 server to serve the content to the client directly rather than redirect the client to a GRM agent cache close to the client.

However, it is more efficient for a GRM L7 server to redirect requests for content that requires a significant amount of bandwidth (for example, PDF, PPT, DOC, and XLS files) to GRM agent caches close to clients. You use the HTTP Redirection Suffix option on the GRM Server - General page in the Appliance Manager to control the HTTP file extensions for which the GRM L7 server is to redirect content. By default, requests for content with .PDF, .PPT., and .XLS extensions are redirected to GRM agent caches close to the requesting clients. HTTP content that does not match the listed extensions to be redirected is served by the GRM L7 server.

Note
GRM preprends the GRM agent cache IP address to the regular URL before sending the 302 response to the client. If a user bookmarks the URL for a document served by the GRM agent cache (for example, a URL for a .PDF file) and the cache becomes unavailable, the bookmark will no longer function. This occurs for HTTP content only.

Rewriting of requests for streaming content

The GRM L7 server redirects all streaming traffic.

Protocol

How rewriting occurs

RTSP

A GRM L7 server sends a 302 response to redirect a client request, and rewrites the URL as

rtsp://redirected-agent-ip/original_url

MMS

A GRM L7 server rewrites the ASX file with the IP address of the closest GRM agent cache, in the following form

Original request:
http://grm-domain/path/file.asx

Contents of the ASX file:
mms://server/path/file.wmv

ASX rewrites the content request as
mms://redirected-agent-ip/server/path/file.wmv

Library Table of Contents Index Previous Next Comments