Increased throughput across data network interface by...

Details Increased throughput across data network interface by... Increased throughput across data network interface by...

1999-05-04

2004-03-02

Nguyen, Steven H. D (Department: 2665)

Multiplex communications

Data flow congestion prevention or control

Flow control of data transmission through a network

C370S230000, C370S477000, C709S225000, C709S229000, C714S748000

Reexamination Certificate

active

06700871

ABSTRACT:

BACKGROUND OF THE INVENTION
A. Field of the Invention
This invention is related to the field of telecommunications and the process by which digital data is transmitted between computer systems over a telephone network or other communications link and a high speed computer network. The invention is particularly suitable for use in devices that support the forwarding of packetized data between two communications media where there is a significant speed differential between the two media. An example of such a device is a network access server (sometimes referred to in the art as a remote access server), where Internet Protocol (IP) or Internet Exchange Protocol (IPX) packets are exchanged between a high-speed local or wide area packet-switched data network and a low-speed dial-up Point-to-Point Protocol link, such as the public switched telephone network. The remote access server connects a remotely located personal computer or other terminal dialing in over a slow PSTN link to a computer or other source of data resident on the high-speed data network.
The present invention achieves an increase in throughput across the interface by dropping redundant packets, that is, dropping the packets entirely or, alternatively, dropping the (redundant) data present in packets going across the interface and transmitting the header for the redundant packet. By dropping redundant packets in the network access server that are en route to the remote terminal, in accordance with a principal aspect of the present invention, the speed and efficiency by which the remote terminal can download files or data from the high-speed network is improved since precious bandwidth on the low-speed link is not being wasted by transmission of packets that have already been received by the remote terminal. The invention can be used in many different application, such as Internet or corporate network access for users dialing in to a network computer over low speed links, e.g., voice grade telephone circuits.
B. Description of Related Art
In order for communication between two computers over a communications medium to be possible, the computers and the equipment in the medium connecting the computers together must follow certain rules or procedures, known in the art as protocols. The communications industry has standards bodies that adopt protocols to govern many different aspects of data communication. These protocols can be modeled as a hierarchy of levels, sometimes referred to as the Open Systems Interface (OSI) model. The lowest level concerns the physical medium connecting the computers together. Above this layer in the model are protocols related to media access control, data link, network, transport, session, presentation, and application features, in ascending order.
Transport layer networking protocols use a variety of techniques to achieve reliable delivery of data. All rely on some form of acknowledgement and retransmission paradigm of some sort. The most common method is to utilize a sliding window protocol with some form of end-to-end acknowledgement required to advance the transmitter's window. Intermediate nodes between the two computers (such as routers) do not participate in the algorithm as they typically provide only network layer services (packet forwarding).
For example, a transport protocol known as Transmission Control Protocol uses a scheme where the receiving endpoint transmits a cumulative acknowledgment of all contiguous data properly received, as well as the currently allowed receive window size (which varies over time), to the transmitting endpoint. The transmitting endpoint uses this feedback information to select what data to (re)transmit. This scheme provides flow control as well as reliable transfer. TCP uses an adaptive retransmission algorithm that is driven based on estimated round-trip times for acknowledgments.
For unicast transport-level flows, packets are generated on the source node, passed through various intermediate nodes, e.g., routers, and finally arrive at the destination. In packet switched networks (such as the Internet), packets can arrive at the destination in order, out of order, more than once, or not at all. The same is true for intermediate nodes. There are two major differences between the intermediate and the destination nodes of a flow. Intermediate nodes are not typically notified of flow establishment/termination, and intermediate nodes do not generally generate feedback information to the source node of the flow. The function provided by most intermediate nodes is to simply forward packets.
In the present state of the art, advanced intermediate nodes are capable of grooming/shaping traffic based on dynamic network load and quality of service (QOS) criteria. This grooming/shaping function consists of reordering packets and selecting packets to drop. Basically, the grooming/shaping function is a scheme to arbitrate the use of congestion points in the network and to provide certain users increased throughput, lower latency, or higher reliability. These types of functions are usually deployed in networks that are over-provisioned. Networks maintained by Internet Service Providers (ISPs) are a common example of an over-provisioned network. They sell more network capacity than they have based on statistical usage patterns to generate revenues.
To perform this grooming/shaping function, intermediate nodes must perform some packet queuing. This queuing allows the nodes to deal with “bursty” traffic without loss of packets. The amount of queuing introduced in a router directly effects maximum forwarding latency (transit delay) and its ability to deal with periods of time where use of an interface on the router is oversubscribed. In general, routers attempt to minimize queuing, as reducing the amount of forwarding latency to a minimum is an overriding concern.
Most users connect to the Internet at the network layer via low-speed modems (<128 Kbps) using the Internet Protocol version 4 (iPv4), encapsulated by the Point To Point Protocol (PPP). The user's phone calls are routed from the public switched telephone network onto the ISP networks via a device known as a Network Access Server (NAS). A representative network access server is described in the patent to Dale M. Walsh, et al., U.S. Pat. No. 5,528,595, which is incorporated by reference herein. The NAS is connected via a high-speed LAN or WAN interface to the Internet and to the user via a relatively low speed PPP modem connection over the public switched telephone network, cellular telephone network or other communications medium. Network access servers similar in architecture and functionality of the above-referenced Walsh et al. patent are currently commercially available from. 3Com Corporation, the assignee of the present invention, and from other vendors in the industry.
It is a widely recognized truism that a user's desired bandwidth for network access (e.g., Internet or corporate network access) is almost always greater than the available bandwidth provided by the modem link. It is also a truism that for most remote access applications, including Internet access, the flow of information and data from the computer on the network to the remote terminal or user is typically much-greater than the flow of data going in the opposite direction. While asymmetrical communications techniques, such as the V.90 56 K technology, help somewhat, this phenomenon usually results in a significant queuing of data directed from the high-speed LAN/WAN interface towards the low-speed modem interface. Moreover, the PPP/modem connection linking the remote user or terminal to the network access server is the bottleneck limiting bandwidth in most remote access scenarios, since the maximum speed of dial-up network access servers is 56 Kbps over a conventional phone line (asymmetrical digital subscriber line (ADSL) and other less common techniques excluded), whereas the transmission rates on the data network is order of magnitudes faster.
Numerous approaches have been proposed in the art to address this bandwidth

Affiliated with

Also associated with

Rating

Increased throughput across data network interface by... does not yet have a rating. At this time, there are no reviews or comments for this patent.

If you have personal experience with Increased throughput across data network interface by..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Increased throughput across data network interface by... will most certainly appreciate the feedback.

Rate now

Comments { 0 }

Profile ID: LFUS-PAI-O-3222419