Abstract: A number of improvements in network adapters that offload protocol processing from the host processor are provided. Specifically, a mechanism for improving connection establishment in a system utilizing an offload network adapter is provided. The connection establishment mechanism provides the ability to offload connection establishment and maintenance of connection state information to the offload network adapter. As a result of this offloading of connection establishment and state information maintenance, the number of communications needed between the host system and the offload network adapter may be reduced. In addition, offloading of these functions to the offload network adapter permits bulk notification of established connections and state information to the host system rather than piecemeal notifications as is present in known computing systems.

Abstract: One embodiment of the present method and apparatus for anonymous data transfers between first and second endpoints in a network comprises forwarding a message through the network, where a default value in the message's time to live field has been modified by an amount such that intermediate nodes or endpoints receiving the message can not infer an ultimate source (e.g., the first endpoint) of the message.

Abstract: One embodiment of the present method and apparatus for anonymous data transfers comprises connecting first and second network endpoints to at least one relay node and transferring data from the first endpoint to the second endpoint through the at least one relay node such that the first and second endpoints are not aware of each other's identities, e.g., are not aware of an ultimate source or destination of transferred data. In further embodiments, an information field specifying a number of times that a data transfer message (e.g., a request, response or get message) should be forwarded is altered so that no receiving node can inferentially identify an originating node.

Abstract: A number of improvements in network adapters that offload protocol processing from the host processor are provided. Specifically, mechanisms for handling memory management and optimization within a system utilizing an offload network adapter are provided. The memory management mechanism permits both buffered sending and receiving of data as well as zero-copy sending and receiving of data. In addition, the memory management mechanism permits grouping of DMA buffers that can be shared among specified connections based on any number of attributes. The memory management mechanism further permits partial send and receive buffer operation, delaying of DMA requests so that they may be communicated to the host system in bulk, and expedited transfer of data to the host system.

Abstract: A number of improvements in network adapters that offload protocol processing from the host processor are provided. Specifically, an improved mechanism for handling receipt of data packets in a system utilizing an offload network adapter. The offload network adapter may include logic that permits the offload network adapter to delay notification of data reception to the host system in different ways. The advantage of delaying the notice of data packet reception to the host system is the potential for aggregation of several data packets, which can arrive immediately after the first one, for example, in a single notification. Given a stream with continuous data packet arrival, a value may be set, either statically or dynamically, for notification delay and this value may be configurable for the host system per communication socket.

Abstract: A method, apparatus, and computer instructions for transferring data objects. A request for a data object is received at a communication adapter. A determination is made as to whether the data object is stored in a memory in the communication adapter in response to receiving the request. The data object is transferred from the memory to a receiver of the data object through the communication adapter in which transfer of the data across a bus connecting the communication adapter to the data processing system is unnecessary in response to the data object being present in the memory.

Abstract: In one embodiment, the present invention is a method and an apparatus for failure-resilient forwarding of data over a computer network. In one embodiment, a marker is introduced into the data stream, e.g., at the sending node, and allows, in turn, forwarding nodes and/or receivers to efficiently track data stream reception. The marker functions as a checkpoint for the data transport process, and is identified and indexed at each forwarding node and receiver. Each receiver saves the marker prior to delivering data to an application, thereby designating a point in the data stream at which all preceding data is confirmed to have been delivered to the application. Thus, if a forwarding node fails, the receiver may request stream data from an alternate forwarding node by specifying to the alternate forwarding node to provide data starting from the marker.

Abstract: A method of clustering communication nodes based on network attributes such as network delays and forwarding capacity; on communication interest attributes; and on application attributes such as quality of service preferences/constraints (e.g. end-to-end delay constraints, bandwidth constraints) in providing communications between users and application servers. A multi-attribute communication feature vector is formed. That vector is comprised of network attributes (such as available bandwidth, client location attributes in the IP map), communication interests attributes (client request for content updates, client subscription to specific data items or to a set of proximal data sources in network space or application/virtual space) and quality of service requirements (such as delay and loss constraints is used to from efficient group communication mechanisms for distributed collaborative applications. Then the multi-attribute communication feature vectors are clustered.

Abstract: A method and apparatus for managing communications for collaborative applications. Middleware uses, network, application, and middleware resource information together with user information to facilitate communications between the application resources and the users via a hierarchical control structure that implements a communication overlay tree. That overlay tree is implemented in view of network constraints, and on the users and on their communication constraints. The middleware provides user index identifiers that inform the application that specific users have communication interests in specific parts of the application space. To send data to users that are interested in a specific part of the application space the application sends that data with a list of user index identifiers. The middleware then associates the user index identifiers with individual users and routes sent data to the individual users along the overlay tree.

Abstract: A method, computer readable media, and apparatus of hierarchical-based communication session and data distribution management that indexes client's communication preferences and network attribute information. Indexing is used to generate dynamic group membership lists that map into communication groups. Indexing is replicated at control nodes in the network overlay to allow distributed management of group membership. Send/receive operations are decoupled through data distribution and the indexing structure. Senders and receivers register their group communication interests to a parent node in the hierarchy. The session control structure aggregates client interest and dynamically updates replicas at control nodes which are selected according to the changes in registered client interest. The indexing structure has self-managing properties for automatic clustering based on client session and data interests, and dynamic partitioning of the session/data interest attribute space.

Abstract: A method for group communication over a network of processors comprises determining an overlay spanning tree comprising an origin node and at least one receiving node, and controlling a source communication rate to be less than or equal to a bottleneck rate of the overlay spanning tree.

Abstract: In one embodiment, the present invention is a method and an apparatus for bandwidth efficient distribution of content over a computer network. In one embodiment, a server cooperates with clients to reduce the usage of server bandwidth, thereby reducing bandwidth-associated distribution costs. The server controls the way that content is distributed within the system, and the clients/receivers may act as intermediate forwarders of data packets. In one embodiment, this is accomplished by sending data items to a first group of receivers on a packet-by-packet basis using unicast communication. The first group of receivers then sends the packets on to a second set of receivers, and the process continues until the packet is forwarded to all intended receivers.

Abstract: A mechanism for bottleneck avoidance is provided in an intelligent adapter. The mechanism allows the adapter to be used such that host/adapter system throughput is optimized. The bottleneck avoidance mechanism of the present invention determines when the adapter becomes a bottleneck. If certain conditions exist, then new connections are refused so that the adapter can process packets for existing connections. If certain other conditions exist, the adapter may migrate workload to the host processor for processing. These conditions may be determined by comparing memory usage or central processing unit usage to predetermined thresholds. Alternatively, the conditions may be determined by comparing adapter response time to host response time.

Abstract: A method and system architecture to manage system resources in such a way as to provide service time guarantees to various classes of work using the system. In the case that the system resources are not sufficient to provide the specified service time guarantees, a method and system architecture is provided to limit work into the system such that the guarantees are satisfied and the revenue derived from the system is maximized.

Abstract: The present invention provides a method, computer program product, and data processing system for efficiently recovering state and performing failover of a network offload engine. The present invention distinguishes between the hard state and the soft state of a protocol. Hard state is state information that, when lost, leads to incorrect protocol behavior. Soft state is state information that may be lost or become inconsistent without loss of correctness. The present invention ensures correctness by always being able to recover the hard state of the protocol. A preferred embodiment of the present invention performs a failover of a network offload engine by temporarily blocking the reception of network packets, recovering hard state from host information, resuming network operation using a substitute network offload engine, and recovering soft state from the subsequent network activity.

Abstract: A method, apparatus, and computer instructions for transferring data objects. A request for a data object is received at a communication adapter. A determination is made as to whether the data object is stored in a memory in the communication adapter in response to receiving the request. The data object is transferred from the memory to a receiver of the data object through the communication adapter in which transfer of the data across a bus connecting the communication adapter to the data processing system is unnecessary in response to the data object being present in the memory.

Abstract: Several deterrence mechanisms suitable for content distribution networks (CDN) are provided. These include a hash-based request routing scheme and a site allocation scheme. The hash-based request routing scheme provides a way to distinguish legitimate requests from bogus requests. Using this mechanism, an attacker is required to generate O(n2)amount of traffic to victimize a CDN-hosted site when the site content is served from n CDN caches. Without these modifications, the attacker must generate only O(n) traffic to bring down the site. The site allocation scheme provides sufficient isolation among CDN-hosted Web sites to prevent an attack on one Web site from making other sites unavailable. Using an allocation strategy based on binary codes, it can be guaranteed that a successful attack on any individual Web site that disables its assigned servers, does not also bring down other Web sites hosted by the CDN.