Multiplex communications – Data flow congestion prevention or control – Flow control of data transmission through a network
Reexamination Certificate
1998-06-09
2001-02-06
Pham, Chi H. (Department: 2731)
Multiplex communications
Data flow congestion prevention or control
Flow control of data transmission through a network
C370S252000
Reexamination Certificate
active
06185189
ABSTRACT:
FIELD OF INVENTION
The present invention relates to high speed data networking and, more particularly, to the interconnection, for bandwidth allocation and control, between Terminating Equipments (TE) and Frame Handlers (FH) in a Frame Relay (FR) switched network by adapting the committed information rate (CIR).
BACKGROUND ART
The present invention makes references to the following standards:
ANSI T1.617, T1.618 and T1.606 (addendum); and
ITU-T Q933 and Q922
The terminology that is used throughout the description is as follows:
FR:Frame Relay is dedicated to High speed switched networks.
FH:Frame Handler concerns the function of FR layer
2
that handles frame switching in the network.
TE:Terminating Equipment is a gateway from FR, it handles upper layer
2
and layer
3
functions.
Bc:Committed Burst size is the number of bytes that can be sent in a burst according to Committed Information Rate.
Be:Excess Burst size is the additional bytes to Bc that may or may not be carried by the network.
Tc:Period over which Bc is sent, its value is generally computed according to the formula CIR=Bc/Tc.
CIR:Committed Information Rate which represents a bit rate that is subscribed from a carrier.
V_CIR:Variable CIR which is the classical recommendation according to the standard organizations as mentioned above.
A_CIR:Adaptive CIR is the main object of this invention.
ARB:Adaptive Rate Base represents the ability to perform flow control by adapting a transmission rate.
PVC:Permanent virtual channel is a leased end to end path through the FR network.
DLCI:Data Link Control Identifier which identifies a PVC over a given hop.
ECN:Explicit Congestion Notification where a bit is piggy-backed in FR Q922 header to notify of a congestion (which happens generally in the FH function). The ECN uses the pair FECN/BECN of congestion management.
FECN:Forward ECN. Generally set by Frame Handler function in a frame that passes through a congested node and sends a signal to the receiving or destination FRTE, advising it to slow down the receipt of information.
BECN:Backward ECN. Flows on the way back from which the congestion was experienced to the source FRTE of the traffic in order to pace it down.
Bottleneck:The node or the line in the network that limits the throughput over a PVC. Both the bottleneck location and its throughput are subject to change overtime.
LAN:Local Area Network, it interconnects stations over a single plant (campus). Tendency is to go from the old shared media model to a switched model to have higher throughput.
WAN:Wide Area Network, everything from the most common modem in a PC to a Serial Optical Network. It interconnects stations and LANs over long distance serial links.
Frame Relay technology has past succeeded over the past years in becoming the major multiprotocol (IP for intra/internet, SNA, SNAP . . . ) Wide Area Network (WAN) for interconnecting Local Area Networks (LAN) of a company's dispersed sites.
Frame Relay is a switched multiplexed technology, which means in other words that many end to end communications can share a given node or a given link without knowing each other.
The load of that given node or link may thus change dramatically over time when huge data exchanges are asynchronously started. Preventing overuse of network resources is thus a major issue. Frame Relay uses Explicit Congestion Notifications to let FRTEs know about a congestion somewhere over a PVC path. The FRTE is responsible for diminishing its throughput till the congestion in the network terminates, through flow control operations.
The definition of objectives and requirements for congestion management is provided in addendum 1 of ANSI T1.606. This standard defines speed and burstiness and describes how the network and the FRTE devices handle an overabundance of data traffic. Congestion occurs when traffic arriving at a resource exceeds the network's capacity. It can also occur for other reasons (equipment failure etc.). Network congestion affects the throughput, delay, and frame loss experienced by the FRTE. The latter should therefore reduce its offered load in the face of network congestion.
When links or nodes in the network are overused, logical bottlenecks tend to form and move throughout the network. A logical bottleneck becomes the place where most of the outstanding data get queued, increasing latency and causing memory storage where it is most unwanted and expensive. When network access speed increases, larger windows are needed and more multiplexing takes place. At some level, the bottleneck cannot keep up and starts discarding frames, which in turn reduces traffic dramatically.
The movement of data through a network sometimes needs traffic signs to indicate when it should stop and go. Access signaling described in the ANSI T1.617, CCITT Q.933, specifies a protocol for establishing and releasing switched frame relay virtual calls and provides a means to inform users of permanent virtual circuits of failure and restoration.
Explicit Congestion Notifications (ECN) by the network are generally a reactive process intended to protect the network nodes. Forward ECN (FECN) and Backward ECN (BECN) are the most common congestion management signals.
Adaptive Rate Based (ARB) protocols are the modern solution to replace window based protocols. Whereas the window based protocols transfer a number of frames and then wait for an acknowledgement, the ARB protocols send a continuous binary flow at a given rate.
Frame relay networks provide a Committed Information Rate (CIR) that is a kind of rate enforcement, refer to the ITU-T X36, but the CIR is still not adaptive. The CIR is the rate (expressed in bits per second) at which the data are transferred between the user equipment and the network.
CIR standard provides a speed range between the committed and excess rates that the terminating equipment can use depending on whether the network reports a congestion or not. In order to adapt to the dynamics of transient logical bottlenecks, the Committed Information Rate should be able to vary within that range. Any implementation that uses a static CIR is bound to be obsolete.
Moreover, in the existing implementations, the ECN are used in a reactive way, meaning that the FECN and BECN are set when a congestion has already occurred. Should the occasion arises, the network enters a slow down condition to recover. In consequence, it is not possible to obtain a fine tuning of the rate at the end point (FRTE). This is why the congestion notification should be issued as soon as the conditions are reached that would lead to a congestion, but before it creates any damage.
Existing implementations try to create a Variable CIR but:
There is no convergence in the algorithm,
A Terminating Equipment (TE) cannot achieve any rate tuning by itself. A cooperation between the switches (FRFHs) AND the end points (FRTEs) as a system is required to achieve a real rate tuning.
The problem addressed here is how to improve that current CIR technique in order to turn it into a rate control and to optimize the throughput to the network to match the logical bottleneck capacity.
OBJECT OF THE INVENTION
The main object of the present invention is to improve the Committed Information Rate (CIR) functionality in order to turn it into an Adaptive Rate Base (ARB) mechanism, so as to apply it to frame relay networks.
The solution is based on the interaction between a proactive setting of the Explicit Congestion Notifications (ECN) in the Frame Handler (FH) function of the switches and a converging Adaptive CIR algorithm in the Terminal Equipments.
The result of this interaction is that when a logical bottleneck is in the process of settling in a switch, the CIR at the Terminating Equipment adapts itself to the throughput of the forming logical bottleneck. Henceforth, the logical bottleneck is exported at the boundary of the network, within the Terminating Equipment.
The end result is that the information sent by the data link control in the Terminating Equipment is paced so that the output matches that of the weakest poi
Brassier Rene
Esteve Denis
Maree Jean-Pierre
Thubert Pascal
Cockburn Joscelyn G.
Frisone John B.
International Business Machines - Corporation
Pham Chi H.
Yao Kwang B.
LandOfFree
Method for adapting committed information rate over frame... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method for adapting committed information rate over frame..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method for adapting committed information rate over frame... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2558353