[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: known request routing
- To: cdn@ops.ietf.org
- Subject: Re: known request routing
- From: Eric Dean <eric@crystalballinc.com>
- Date: Thu, 19 Apr 2001 09:31:54 -0400
- Delivery-date: Thu, 19 Apr 2001 07:26:41 -0700
- Envelope-to: cdn-data@psg.com
Comments:
Section 2.8.5: There are some implementations of DNS that forward the client's IP
address within the DNS packet's "Additional" section.
Section 4.2.3: Raises security concerns and may have problems with zipped/encrypted
content.
I also notice that there are "Pure" DNS-based request routing methods and then there
are those that have a DNS+URL based method. A distinction is that a Pure DNS method
allows for caching surrogates to run in a native mode for it's internal and peered
CDNs. By this, I mean that it can serve content regardless of the origin server.
A DNS+URL based mechanism that encodes specific CDN control information after the
FQDN (in what would appear to be the absolute path), such as http://cdn.foo.com/cgi-
bin/foo/img.gif... would have difficulties interoperating with other caching
surrogates that apply only native request logic.
Section 6.2.1: You may also want to highlight that probing techniques often only
probe the client DNS server and not the client. Also, while probing only occurs
periodically, client DNS servers remain relatively immobile.
--
Eric Dean
President, Crystal Ball Inc.
W 703-322-8000
F 703-322-8010
M 703-597-6921