The present invention relates to content request routing and is particularly concerned with routing between content network peers.
A current Internet draft “Known Mechanisms for Content Internetworking” by F. Douglas et al. provides a useful overview of content internetworking. (Nov. 8, 2001) The use of “Intelligent DNS (IDNS) is discussed as a way to handle request redirection.
Content networks play an important role in the overall architecture of the web. In the Internet today several approaches have been proposed for providing infrastructure, at layers 4 through 7, to get content to end users or user agents in a scalable, reliable, and cost-effective fashion. In this regard, various protocols and appliances have been developed for the location, download, and usage tracking of content. Examples of such technologies include: web caching proxies, content management tools, and intelligent web switches.
In general, a content network can be viewed as a virtual content overlay network in the OSI stack. This content overlay layer enables the delivery of richer services that rely on underlying elements from all 7 layers of the stack to subscribers or end users. Content overlay services rely on layer 7 protocols such as HTTP or RTSP for transport.
However, regardless of the size of a content network, its ability to serve clients and subscribers is limited by economic realities and other factors. Hence, in order to increase the scale, reach and performance of content networks, it is possible to interconnect them. Hence, content internetworking is the interconnection of multiple content networks. In the networking field, content internetworking is also known as content distribution internetworking or content peering.
Content internetworking allows different content networks to cooperate to serve content to end users or user agents. This leads to the ability of content networks to share resources so as to provide larger scale and/or reach to each participant than they could otherwise achieve. In order to be able to interconnect content networks, various architectural components must be introduced. In particular, the interconnection of content networks is achieved through the establishment of common internetworking gateway. The gateway must be able to provide mechanisms to distribute content or inject content into the networks and must also be able to direct user requests between them. The task of directing users requests to various surrogates among the inter-networked content networks is also called request routing and it is done in a content router. It is also possible to define various accounting and authorization schemes for financial settlements. In the literature, a content router is also called a request routing system (RRS).
There are various request routing techniques that could be used within a content router to direct users request for content among internet-worked content networks. At a high-level, these may be classified under: DNS request-routing, transport-layer request-routing, and application-layer request-routing. However, regardless of the technique that is used for request routing, there should exist mechanisms within the content router that insures that the task of directing users requests is performed in a loop free manner.
Among the various request routing techniques, the use of DNS based mechanisms has gained popularity due to the ubiquity of DNS as a directory service. In DNS based request-routing techniques, the content router acts as a specialized DNS server that is inserted in the DNS resolution process. The content router is capable of returning a different set of A, NS or CNAME records based on defined policies, network conditions and cost. Such DNS servers have also been called Intelligent DNS servers (IDNS). In general, within IDNS, the use of CNAME redirection techniques is the preferred method of request routing.
Referring to
The IDNS brokering server 20 uses CNAME or NS redirection to other CDNs, or it can forward the DNS request directly to a CDN that will respond to the request directly (not shown in
Request-routing systems (RRS) present a “black-box” view of their associated distribution systems. Since in such an environment no CDN possesses a global view of all other CDNs, the request-routing system has to rely on a peer-to-peer model in which each request-routing system is only aware of its direct neighbor.
There are two known methods for redirecting a request between two interconnected request-routing systems. The first method is an interactive method where a RRS directs the request to the next-hest (neighbor) RRS. This continues until a surrogate is finally selected. The second method is recursive where a RRS directs a request to the next-best RRS but expects an answer to return to the client. These two methods are analogous to recursive vs. iterative DNS lookups.
The interactive approach will either find a CDN that will accept the request or it will return a request failed message. With either result, considerable resources are involved with the messaging that is exchanged in this process.
Referring to
A second method of request redirection is recursive. Here the RRS redirects a request to the next best RRS, but expects the new RRS to return a reply to the client. Either of these methods will work if there is one level of peering. However, if there are multiple levels of peering, these simple redirection schemes may lead to looping, where the request continues to be past among the CDN peers without converging on a CDN that will accept the request.
An object of the present invention is to provide an improved content request routing method.
Accordingly, the present invention defines peering levels at distribution of content from a content source then tags each redirection of requests with an identifier of the CDN forwarding the request to a peer. Advantageously, the peers only require knowledge of nearest neighbours.
Conveniently, a content by CDN matrix is populated at the time of distribution.
a, 4b, and 4c illustrate content distribution tables for nearest peers in accordance with an embodiment of the present invention; and
Referring to
In operation, a client 130 sends a request 132 addressed {www.a.com/α} for content α to CDNA 60. However, CDNA 60 does not currently have the capacity to deliver this content. So it refers to its content distribution table 134 to see if the neighboring peers are able to provide this content. The CDNA 60 sees that it has a level I peer, CDNB 70 that has the α content and forwards the request to CDNB 70 addressed (www.a.b.com/α).
The CDNB 70 is also too busy to handle the request and determines from its distribution table 138 that its level 1 peer CDNC 80 does not have the content α, however its level 2 peer CDND 90 does. In this case, the CDNB 70 forwards the request 140 addressed, www.a.b.d.com/α, to the CDND 90. The CDND 90, is also too busy to handle the request so consults its distribution table 142 and determines that the only two peers having content α are CDNA 60 and CDNB 70, but the address ww.a.b.d.com/α shows that these two peers have already refused the request. At this point CDND 90 returns a network busy request refused message 144 to the client 130.
Referring to
In the example above, the initial CDN peer was CDNA. If the CDN peer had been CDN B, the addressing of the re-routed request would be www.b.nextpeer.com. Hence, the re-routed request has the general form:
Referring to
This application is a continuation of co-pending U.S. patent application Ser. No. 10/013,677, entitled “Content Request Routing Method,” which was filed on Dec. 13, 2001, and which is hereby incorporated herein by reference in its entirety.
Number | Date | Country | |
---|---|---|---|
Parent | 10013677 | Dec 2001 | US |
Child | 13860092 | US |