Cisco Technology, Inc.

Massachusetts Institute of Technology

Barbara Liskov

John F Carr

MArk Stuart Day

Stephen Jeffrey Morris

James O'Toole

Dmitry Stavisky

Miguel Castro

Kirk Johnson

M Frans Kaashoek

Gang Lu

Dan Li

David M Bornstein

Content servers are directed to send messages to network nodes which are used to obtain IP addresses and to receive responses from the network nodes. The responses provide network service information. The content servers collect data based on the responses and feed the data to a routing server. Based on the data fed to the routing server, respective choices of content servers are selected and provided to the network nodes to respond to clients requesting content server addresses from the network nodes.

Content server selection for accessing content in a content distribution network

A content engine receives a content request from a client, generates a result including one of a content-providing value and a redirection value in response to the content request, and selectively provides, to the client, one of (i) content when the result includes the content-providing value, and (ii) a redirection message when the result includes the redirection value, the redirection message including an extended domain name having a client identifier which identifies the client. The result can be used by a content router or content engine in deciding whether to provide content or a redirection. The decision to selectively provide can be based on client proximity to the content engines, past redirections, content size and type, content engine capacity, content availability and other factors.

Methods and apparatus for redirecting requests for content

Mechanisms and techniques operating in a content distribution network in a peering relationship with a second content distribution network. The system maintains a first client list identifying a first set of client devices associated with a first content distribution network. The system also obtains a second client list identifying a second set of client devices associated with a second content distribution network. The system applies a client list filtering technique to the first client list and the second client list to produce a preferred client list associated with the second content distribution network. The preferred client list potentially identifies at least one preferred client device for which the second content distribution network preferably processes client requests. The system then operates a content routing technique to direct a client request associated with a client device to a content distribution network having a preferred client list that identifies the client device associated with the client request as a preferred client device.

Methods and apparatus for processing client requests in a content distribution network using client lists

The invention is directed to techniques in a computer, for processing a content request. In one arrangement a content engine receives a content request from a client, generates a result based on the content request, the result including one of a content-providing value and a redirection value in response to the content request, and selectively provides, to the client, one of (i) content when the result includes the content-providing value, and (ii) a redirection message when the result includes the redirection value, the redirection message including an extended domain name having a client identifier which identifies the client. The result can be used by a content router or content engine in deciding whether to provide content or a redirection. The decision to selectively provide can be based on client proximity to the content engines, past redirections, content size and type, content engine capacity, content availability and other factors.

A network of nodes caches replicated datasets in which dataset changes are efficiently propagated as a set of changes even under failure conditions. A master node and a plurality of subordinate nodes in the network each maintain a copy of the dataset and a change log storing change events in the dataset in that node. The change log further includes a rename chain having a plurality of linked rename records created in response to a new master gaining control of the dataset. The master node computes and propagates dataset changes to the subordinate nodes as a set of change events. If the master node fails, one of the subordinate nodes becomes temporary master and continues to propagate dataset changes using its dataset and its change log in response to update requests from other nodes where the update request contains information from the change log of the requestor node.

Method and apparatus for efficient propagation of large datasets under failure conditions

In a network having a plurality of routers and also an application running at a plurality of replicas, a system for maintaining a client session between a client and a particular replica includes providing a database of bindings of client request identifiers to replicas. The system further includes a change log of changes made to the database where each entry has an event number. A version vector associated with the database and the change log holds a most recent event number for each router in the network. The system receives updates of change events made at the routers in the network and reconciles the updates with the database and current version vector. The system has data to maintain a client session between the client and the particular replica through repeated requests. Also the system and routers share data so that the client session can be maintained no matter which router handles subsequent client requests.

Method and apparatus for providing sticky bindings using version vectors

A technique can be used to obtain content (e.g., a live feed, pre-positioned content, etc.) from a content-originating device (a content source). The technique involves identifying a tree-based location-path having a series of locations which leads from the computerized device to the content-originating device. Each location includes a set of devices, and the set of devices of at least one location includes multiple devices. The technique further involves selecting a device-path from the computerized device to the content-originating device based on the identified location-path, and acquiring the content from the content-originating device from at least one of the devices along the selected device-path. The selected device-path includes at least one device of each location of the series of locations.

Managing a content distribution system using virtual content-distribution trees

Conventional replication status reporting techniques in a Content Delivery Network (CDN) tend to produce large, unwieldy reports which are difficult to transmit, consume large amounts of bandwidth, and require a large share of CPU resources to process. A replication status reporter which identifies an ordered list of events corresponding to content items (files) for replication (pre-positioning) in a content delivery network, and determines a point of progression along a sequence represented by the ordered list of files, allows a management server to avoid computationally expensive endeavors which itemize and transmit lists of every replicated file in the CDN. The servers in a CDN channel receive files as a series of ordered events. Origin servers deliver the files generally according the event. A replication status report indicates the progression in the list, rather than enumerating the entire list, avoids cumbersome and unnecessary processing and delays in reporting replication status.

Content delivery network (CDN) replication status reporter

Mechanisms and techniques provide a system that operates in content routers and content engines in peered content distribution networks. Using domain name specifications associated with the content distribution networks, a content router may provide a forward domain resolution response to redirect a domain resolution request to another peered content distribution network. If a secondary content router determines that a content request is not to be processed in its content distribution network, it may respond with: a reflected domain resolution response that is returned to a primary content router or a forward domain resolution response. Content engines may also reflect content requests back to the primary content router by providing: a reflected redirect content response to a client, or a redirect forward content response that must be resolved. After termination of a peering relationship, a reflection continuation period may be defined during which secondary content networks provide reflected or forwarded responses.

Methods and apparatus for processing content requests using domain name service

Mechanisms and techniques provide a system that operates in a computerized device(s) to establish a data distribution path for content within a content distribution network by receiving a network topology definition defining at least one hierarchical interconnection of network groups. Each network group comprises at least one content engine. The system also receives a channel definition comprising a selection of a plurality of content engines that are to distribute content within the content distribution network. The plurality of content engines in the channel definition is selected from content engines within the network groups defined within the network topology definition. The system determines an assignment of at least one root content engine within the channel definition and applies a content distribution path determination technique to the network topology definition in relation to the channel definition to determine a set of content distribution paths in the content distribution network to be used for distribution of content from the root content engine(s) to the content engines defined in the channel definition.

Methods and apparatus for performing content distribution in a content distribution network

Methods and apparatus for obtaining content from a content-originating device within a computerized network

A network operator identifies an address of a network element used by a particular client to obtain IP addresses. The client sends a test message to a test URL which includes a unique host name unknown to the client or network element. The network element accesses an authoritative server, which records the IP address of the requesting network element and resolves the host name in the test URL to a test IP address. The server records an IP address of the client when the client subsequently sends a message directly to the test IP address. In an alternative embodiment, another network node generates a redirect command, which incorporates the client IP address in a test URL, to allow for single-message determination of the network element and client IP addresses by the authoritative server. In both embodiments, user unique test URLs may be generated to allow the network operator to assist many users.

Remote identification of client and DNS proxy IP addresses

A new approach for asynchronous state-machine replication in a fault-tolerant system offers both integrity and high availability in the presence of Byzantine faults. The approach also improves the security of previous systems by recovering replicas proactively without necessarily identifying that they have failed or been attacked. This proactive recovery limits the time extent of a particular fault by regularly recovering replicas. In this way, the system works correctly even when all the replicas fail multiple times over the lifetime of the system, provided that less than &frac13; of the replicas are all faulty within a window of vulnerability. The approach also features an efficient implementation of message authentication that prevents an attacker from impersonating a replicated node that was faulty after that node recovers

Growing community of inventors

Waltham, MA, United States of America

Barbara Liskov