If a center or edge cache to which a client sends a request for CDN content does not have the requested content (that is, when a cache miss occurs), the cache fetches it from the CDN content servers directly. You must determine the method to use for changing (rewriting) the host name in client requests so that looping does not occur with cache misses for CDN content. When host name rewriting is set up, any upstream transparent caches that intercept the request will not redirect traffic back through the CDN. Therefore, looping is avoided. This looping would occur, for example, if a Web accelerator were deployed in front of an enterprise's origin server.
Available methods are
These methods are applicable for deployments using GRM DNS servers, GRM L7 servers, or both.
Note
If the GRM DNS server takes over a content provider's existing DNS, do not use accelerator rules. If you do so, you will duplicate your effort.
A CDN hoster can configure its DNS nameserver to return a canonical name (CNAME) with the associated IP address. Returning the CNAME along with the IP address avoids looping that might occur otherwise. The cache can then fetch content from the origin server by using the canonical name and store the content under the original name.
For example, assume that the CDN hoster asked a content provider to rewrite URLs to images.enterprise.cdnhoster.com. No origin server actually exists that can serve content for images.enterprise.cdnhoster.com if a cache must resolve a cache miss. Therefore, the CDN hoster adds to its DNS database a CNAME record, images.enterprise.com, that corresponds to the fully qualified domain name (FQDN) images.enterprise.cdnhoster.com. Subsequently, when a cache miss occurs, the cache uses the CNAME to rewrite the host header and fetches the content from the content provider's origin server (DNS CNAME rewriting).
Using NetCache accelerator rules for host-name rewriting eliminates the need for the DNS nameserver that can rewrite the host name and configuring DNS CNAME rewriting. On each center or edge cache, you must configure standard NetCache accelerator rules with origin server addresses to force the cache to serve content from the origin servers only.
A disadvantage of using accelerator rules is that this method is not very scalable.
If you are using accelerator rules, Network Appliance recommends that you use the NetCache multiple-cache management feature. Multiple-cache management simplifies distribution of the accelerator rules among the caches.
Copyright © 1994-2002, Network Appliance, Inc. All rights
reserved.
Part No. 210-00374
File: grmcont3.htm
Last updated: 11/12/02 17:45:13