Abstract: A method for sending data over a network from a host computer. The host computer includes an operating system comprising at least a user space and a kernel space. The amount of data provided from the user space to the kernel space within one system call exceeds the size of an IP packet. A loop function in an application in the user space sends multiple packets to the kernel space within a single system call containing IO vectors which contain pointers to the data in the user space. A last data unit being processed may be designated using a flag included in the message header. In the kernel space a second loop function is used to reassemble the vector groups and pass them down the network stack. The data may then be passed to the network hardware using a direct memory access transfer directly from the user space to the network hardware.

Abstract: A method and apparatus for file access in a storage access network (SAN) wherein a portion of a first file is accessed and an access rate computed based on time and amount of data accessed. If the computed access rate is less than an expected rate, a problem is identified and analysis performed. A first analysis accesses the first file through a second link to determine if the link is the problem. If the first file is accessed at an expected rate, the first link is identified as the problem and the file is accessed through an alternate link. A second analysis accesses a second file through the first link. If the second file is accessed at an expected rate, the first file is identified as the problem and the second file is accessed through the first link.

Abstract: In the method for management of data transmissions in a network, switching means are provided, by means of which a change can be carried out between a first data transmission type and a second data transmission type, as alternative data transmission types between a transmitter and a receiver. The change is carried out on the basis of at least one respectively predetermined criterion. The first data transmission type corresponds to a transport protocol based on rate-based overload control. The second data transmission type corresponds to a transport protocol based on RTT-based overload control.

Abstract: The invention relates to a method for transporting data over a data connection between a sending network component and a receiving network component via a network. The method solves a problem with the reliable data transport protocol according to which a number of data transport packets will be acknowledged in regular time intervals. When such an acknowledge message is lost, the sending network component will have to keep the data for the purpose of retransmission and cannot free its sending buffer. The invention proposes to repeat such an acknowledge message as a precaution for the case that it gets lost after a short time. The time for repeating is set to be less than an RTT (round trip time for the connection). The solution disregards the fact that no statement about the reception of the acknowledge message can get back to the receiver before the RTT has elapsed.

Abstract: The invention relates to a method for managing network resources in a network with a network management device (M), a network component (A), and a further network component (B), and a network management device (M), the method comprising the following steps: transmitting a request for an advance reservation of a network resource by the network component (A) to the network management device (M), wherein the request includes information regarding a reservation start time and a reservation end time; transmitting a request for an ad-hoc reservation of a further network resource by the further network component (B) to the network management device (M); allocating the further network resource to the further network component (B) by the network management device (M); and allocating the network resource to the network component (A) by the network management device (M) at the reservation start time, wherein the network resource comprises a part or all of the further network resource.

Abstract: The invention relates to a method for managing a data connection between a sending network component and a receiving network component via a network, and a network component, the method comprising the following steps: sending an initiating message to the receiving network component by the sending network component, to initiate a data connection between the sending network component and the receiving network component; receiving the initiating message by the receiving network component; sending further messages to the receiving network component by the sending network component; receiving the further messages by the receiving network component calculating a transmission delay for at least one of the further messages received by the receiving network component; estimating optimal transmission parameters for transmitting messages between the sending network component and the receiving network component based on the calculated transmission delay, wherein estimating of the optimal transmission parameters substanti

Abstract: The invention relates to a method for synchronizing a clock of a network component with a clock of a further network component communicatively connected to the network component in a network and a network component, whereby the method comprises the following steps: determining a set of forward transition delays for transmission of messages from the network component to the further network component and a set of corresponding backward transition delays for transmission of messages from the further network component to the network component, selecting a minimum forward transition delay from the set of forward transition delays, selecting a minimum backward transition delay from the set of backward transition delays, deriving an estimated value for an offset between the clock of the network component and the clock of the further network component from the selected minimum forward transition delay and minimum backward transition delay, and adjusting the clock of the network component in accordance with the estima

Abstract: The invention relates to a method of operating a network subnet manager, in particular Infiniband network subnet manager in a subnet of a network. The subnet manager is a software entity with the task of managing, for example Infiniband subnet and can reside on any one node. The subnet manager discovers the topology of the subnet that it manages, assigns a subnet ID to each port, assigns an address to each port in the subnet, establishes the possible path between all end nodes in the subnet, sets QoS parameters and sweeps the subnet on a regular basis looking for topology changes. Only one subnet manager entity can be master within a subnet. A problem related with this specification is that a specific vendor implementation may exist with many features and internal functionality that are not specified in the network specification.

Abstract: The invention relates to a new type of quality of service architecture for a network assembly and a corresponding computer network system. The Internet technology as approved by the IETF organization has defined different services. One is the integrated services (IntServ) and the other the differentiated services (DiffServ) architecture. Another is the RSVP Protocol according to which Quality of Service can be implemented with accuracy and richer functionality. The invention shows a way how less sophisticated QoS managers inside the network with only DiffServ or Intserv capability, can be used for enhancing the QoS functionality based on so-called RSVP shadowing messages, and easy to implement RSVP shadowing stacks.

Abstract: Modern computers (10, 20) come with different timers having different attributes like time resolution, supported time range and time reference. Some are local timers, representing relative time values like the TSC counter (11, 21) counting CPU cycles from the power on or reset of the computer. Some are global timers, representing an absolute or real time. Having different classes of timers in the computer (10, 20), a number of computations need to be performed to tie the different timers to each other. It is the idea of the invention to define a single high resolution timer structure wherein the time value is represented with a digital number and a flag clarifies whether the timer is a global timer or local timer. The invention enables a much faster processing of the time values. The global time, tied to a high-resolution timer can be tracked much faster. This means better precision of scheduling of time critical jobs, like packet scheduling for data transport.

Abstract: The invention relates to a method for managing network resources in a network with a network management device (M), a network component (A), and a further network component (B), and a network management device (M), the method comprising the following steps: transmitting a request for an advance reservation of a network resource by the network component (A) to the network management device (M), wherein the request includes information regarding a reservation start time and a reservation end time; transmitting a request for an ad-hoc reservation of a further network resource by the further network component (B) to the network management device (M); allocating the further network resource to the further network component (B) by the network management device (M); and allocating the network resource to the network component (A) by the network management device (M) at the reservation start time, wherein the network resource comprises a part or all of the further network resource.

Abstract: The invention relates to the field of network communication in a wide area, where a local network of a first type has a sending station that communicates with a receiving station in a local network of a second type. A network of a third type is in between the two networks and provides virtual private networking between the two local networks. The network of the first type supports a fine grained QoS, whereas the network of the third type supports a coarser grained QoS. In one example the network of the first type is RSVP capable and the network of the second type is an MPLS network. The invention resides in a component called RSVP-MPLS proxy that maps the RSVP resource advertisements and reservations within an RSVP-aware customer network to an MPLS network, whereby the receiver side doesn't participate in the RSVP communication process.