Multiplex communications – Pathfinding or routing
Reexamination Certificate
1999-08-20
2003-10-21
Olms, Douglas (Department: 2661)
Multiplex communications
Pathfinding or routing
C370S401000, C370S449000
Reexamination Certificate
active
06636501
ABSTRACT:
FIELD OF INVENTION
This invention deals with an improved high-speed digital network node communication system and method for speeding-up node traffic, and more particularly with a transparent modular processor architecture, particularly useful in high speed digital networks operating in an Internet environment.
BACKGROUND OF THE INVENTION
to Modern digital networks operate in a multimedia environment for transporting different types of data (including pure data, i.e. files representing alphanumeric characters, as well as data representing digitized and encoded voice, image, video signals etc.). The network should naturally ensure compliance with various requirements specific to each kind of data.
Different techniques have been developed for transporting data from one location to another throughout the world. These techniques include packet switching, whereby the data are arranged into so-called packets. Those packets may either be of predefined fixed length, as in Asynchronous Transfer Mode (ATM), or be of variable length, like in Packet Type Multiplexing (PTM) mode of operation, said PTM being currently used for transporting voice data. The basic aim of both packet switching techniques is to allow a multiplexing of the different types of data over transmission links, to optimize, as much as possible, the available transmission bandwidth. Consequently, a large number of networks, both public and private, with possible interconnections, have been developed for transporting those data throughout the world.
On the other hand, the evolution of telecommunication in general, and of packet switching networks in particular, is driven by many factors among which technology evolution and application are worth being emphasized.
As far as technology is concerned, obviously considerable progress has been achieved recently with the maturing of new transmission media, e.g. optical fiber links. High speed rates can now be sustained with very low error rates. For example, very important bandwidth are profitable for long distance networks as well as for high rate local networks. Accordingly, universal use of digital technologies appeared for both private and public telecommunication networks.
With these new emerging technologies, many potential applications that were not possible in the past are now becoming accessible and attractive. In this environment, generic requirements are now expressed by the users, such as:
Improving old applications. Sub-second response times, which are achievable on low-cost personal computers, have raised user expectations so that the slow response times that were acceptable on wide area networks some years ago are no longer tolerable today. The user interface can be made better, for example, with fast response, full screen applications.
Enabling new applications. Emerging applications like graphic, image, video and multimedia processing are generating a fairly large amount of traffic. These new applications that were not feasible (or even thinkable) not too long ago are now accessible and generating an ever-increasing demand on bandwidth.
Optimizing communication networks. There is a need for rationalizing the many disparate networks that major users have. Investments can be optimized by integrating heterogeneous traffic like voice, video, and data over the same transport facilities regardless of protocols. On the other hand, users want the opportunity to control their networking cost by choosing among the different price/performance options offered by the variety of vendors and carriers and to maximize their ability to take advantage of applications built on top of disparate underlying network technologies.
Accordingly, one may have noticed an explosion in demand for high-speed digital network facilities which led to the so-called service providers (like Internet service providers or ISPs) using core backbones offering high-speed data transport facilities to large numbers of heterogeneous users' traffic, possibly through so-called “access backbones” or “core backbones”. Said service providers should be transparent to users and offer fairly large communication bandwidth for lease at low cost.
For instance, service providers are now running, or expect to run, important core backbones and offer their service to users transporting traffic between distant locations throughout the world, which makes particularly sense when the user's traffic includes multimedia.
Accordingly, these core backbones may interface to both high-speed links (e.g. backbone network links) operating between 100 and 1000 Mbps, and lower-speed links operating, say, between 64 Kbps and 50 bps. This leads to traffic aggregation with a requirement for switching high to low rate traffic (and vice versa) in high speed network nodes, quite fast.
As far as this invention is concerned one may schematically envision a high speed network node as including input/output ports throuhg which the data enter/exit the node; a huge memory device for buffering data received from ports, while a node processor dynamically identifies the node exit port toward which said received data should be directed via a node switching device.
For the data being packetized, each packet includes a so-called packet header, and either a fixed length (e.g. 48 bytes) payload in Asynchronous Transfer Mode (ATM), or a variable length payload (e.g. up to 64 Kbytes), in Packet Type Multiplexing (PTM) mode. In both instances the header includes destination identification, and we shall consider that said header packet destination identification may enable defining both the considered node exit port to be used for the packet being processed, and the next network node the packet should be forwarded to, and all this, via simple table look-up.
To that end, the node memory store a so-called “routing table” dynamically filled-in during normal network traffic operation. Then, in operation, while a received packet is being buffered in the node memory, the system, by simply reading the network header and addressing the routing table accordingly, can identify the node exit port and orient the considered packet toward said exit port on a FIFO basis. But it should be noted that a routing table may include 5000 to 15000 entries, and at the high transmission rates actually being practiced, even such a simple table look-up to help matching a packet header indication with a table entry might take too long given the presently existing node processors power, while the node packet transfer should be quite fast and transparent.
This problem was already addressed and several solutions (both software and hardware solutions) have been proposed. Hardware solutions for directly pointing to the right routing table entry matching with any packet header raises some problems due to the fact that not only presently used header formats may vary, but in operation the situation might worsen with standardized header length increasing from 4 bytes long headers up to 16 bytes long headers. In addition, such hardware solutions lack flexibility to enable implementing these in currently existing networks throughout the world.
A number of software-implemented solutions to the problem have also been proposed to speed-up the table look-up operations higher and higher performance algorithms. With these implementations, while any received frame (or packet) would be conventionally buffered, the improved algorithms should enable analyzing the framed header and looking for faster matching with the routing table contents/entry. While one may expect a more flexible implementation of such software-oriented solutions into existing networks, these rather complex solutions would add to the node processor workload, which processor presently often operate in a rather serial mode. Therefore, even said soft implementable solutions to the addressed problem still lack efficiency with presently operating network node architectures.
Accordingly one may say that both available soft and hard solutions to enable faster switching of the data frames between network node entry and exit ports do not
Dispensa Jean-Claude
Esteve Denis
Jourdet Christophe
Thubert Pascal
International Business Machines - Corporation
Nguyen Brian
Olms Douglas
LandOfFree
High-speed digital network node communication system and method does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with High-speed digital network node communication system and method, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and High-speed digital network node communication system and method will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3174143