Explicit rate computation for flow control in compute networks

Details Explicit rate computation for flow control in compute networks Explicit rate computation for flow control in compute networks

1998-12-11

2003-04-15

Yao, Kwang Bin (Department: 2662)

Multiplex communications

Data flow congestion prevention or control

Flow control of data transmission through a network

C370S232000, C370S231000

Reexamination Certificate

active

06549517

ABSTRACT:

BACKGROUND OF THE INVENTION
The present invention relates to flow control in a communications network and, more particularly, to an Explicit Rate (ER) computation for Available Bit Rate (ABR) traffic to achieve flow control in an Asynchronous Transfer Mode (ATM) network.
The
ATM Forum Traffic Management Specification
, Version 4.0, 1996 (TM4.0), specifies five classes of service to support the diverse requirements of multimedia traffic: Constant Bit Rate (CBR), Real-Time Variable Bit Rate (rt-VBR), Non-Real-Time Variable Bit Rate (nrt-VBR), Available Bit Rate (ABR), and Unspecified Bit Rate (UBR). The UBR traffic class provides a best-effort service: No amount of bandwidth is guaranteed, and any cells may be discarded. ABR and UBR are best suited for LAN internetworking and other types of data traffic. UBR is directed at delay-tolerant applications (e.g., file transfer and e-mail). It provides no feedback about network congestion to the user or application. Thus, UBR increases the risk of discarded cells, which in turn increases network traffic because of the lost cells that must be retransmitted.
The ATM Forum defined ABR service to improve service to bursty sources that would instead use UBR. ABR is intended for data applications that can adapt to time-varying bandwidth and tolerate unpredictable end-to-end cell delays. ABR connections share the available bandwidth. The concept of available bandwidth is intrinsic to the service; it is whatever bandwidth exists in excess of CBR/VBR traffic, as defined by the network provider. Thus, the ABR traffic is allowed to use bandwidth that would otherwise be unused, increasing the link utilization without affecting the quality of service (QoS) of CBR/VBR connections. The main practical difference between ABR and UBR is that for ABR, the network provides congestion information to the application. This lets the application constantly modify the transmission rate, achieving the best throughput.
The ABR service class is designed so that applications can grab any unused network resources that VBR and CBR traffic does not utilize, i.e., unused bandwidth and buffer space. Gains due to statistical resource utilization, however, come at the risk of potential congestion when many applications compete for network resources. Therefore, proper congestion control must be in place to ensure that network resources can be shared in a fair manner and that performance objectives such as cell loss ratio can be maintained. In TM4.0, the ATM Forum has specified a flow control mechanism which supports several types of feedback to control the source rate in response to changing transfer characteristics. The purpose of feedback in the context of ABR service is to use available bandwidth (after allocation to CBR and VBR sources) efficiently and allocate it evenly among active ABR connections. Other objectives include instantaneous access to bandwidth which is required to offer dynamic ABR service. This feedback information is conveyed to the source, which adapts its traffic in accordance with the feedback. The feedback information includes the state of congestion and a fair share of the available bandwidth according to a network-specific allocation policy. To ensure interoperability, an ABR end system must always implement the standard-defined source and destination behavior in TM4.0.
The ABR congestion control scheme specified in TM4.0 is a rate-based, closed-loop control that utilizes the feedback information from the network to regulate the rate of cell transmission at the source.
FIG. 1
illustrates the basic operation of ABR congestion control. In network
100
, source
102
transmits data cells
108
to destination
104
via one or more switches in network
100
, one of which, switch
106
, is shown in FIG.
1
. Source
102
also generates special probe cells
110
referred to as forward resource management (RM) cells in proportion to its current data cell rate. The parameter Nrm specifies the maximum number of cells a source may send for each forward RM cell. Thus, source
102
normally sends a forward RM cell
110
for every Nrm-1 data cells
108
. Destination
104
turns around and sends RM cells
110
back to source
102
in the other direction. These cells are referred to as backward RM cells, and are shown as cells
112
in FIG.
1
.
The RM cells, which can be examined and modified by the switches in both forward and backward directions, carry feedback information of the state of congestion and the fair rate allocation. TM4.0, which is incorporated herein by reference, presents a precise definition of the source end system and destination end system behavior, the content and format of RM cells, and a range of feasible switch mechanisms. These mechanisms are characterized by different levels of complexity and achieve varying degrees of fairness. The wide range of options demonstrates the flexibility in the choice of switch mechanisms available with the rate-based framework. With reference to
FIG. 1
, the following summarizes the operation of the rate-based control scheme, the details of which are found in TM4.0.
A switch shall implement at least one of the following methods to control congestion:
a) Explicit Forward Congestion Indication (EFCI) marking: Switch
106
may set the EFCI condition in the header of an ATM data cell
108
(using the payload type field) as it passes in the forward direction. This causes destination end system
104
to set the congestion indication (CI) bit in a backward RM (BRM) cell
112
. Most first-generation ATM switches implemented this mechanism before the RM cell was fully defined.
b) Relative rate marking: Switch
106
may set the CI or the no increase (NI) bit of a passing forward RM cell
110
or backward RM cell
112
. If the bit is set in a forward RM (FRM) cell
110
, that bit will remain set in the corresponding BRM cell
112
. More rapid results are achieved by setting one of these bits in a passing BRM cell
112
. To achieve the most rapid result, a switch may generate a BRM cell
112
with the Cl or NI bit set rather than wait for a passing BRM cell.
c) Explicit rate marking: Switch
106
may reduce the value of the explicit rate (ER) field of an FRM cell
110
and/or BRM cell
112
.
Switches that implement the first two options above are known as binary switches; they can reduce implementation complexity but may result in unfairness, congestion oscillation, and slow response. Switches that implement the last option are generally called ER switches and require sophisticated mechanisms at the switches to compute a fair share of the bandwidth. The standard-defined source and destination behaviors, however, allow interoperation of the above three options.
With reference to
FIG. 1
, source end system
102
sets up a connection with a call setup request for ABR connection. During this signaling setup, values for a set of ABR-specific parameters are signaled by source end system
102
and the network elements. Some of these parameters are requested by source
102
, based on its requirements, and can subsequently be modified by the network (e.g., Peak Cell Rate (PCR), Minimum Cell Rate (MCR)), while others can be set by the network (e.g., those impacting rate increase/decrease behavior such as Rate Increase Factor (RIF), Rate Decrease Factor (RDF), and Nrm). An application using ABR service specifies a Peak Cell Rate (PCR) that it will use and a Minimum Cell Rate (MCR) that it requires. The network allocates resources so that all ABR applications receive at least their MCR capacity. The network then shares any unused capacity in a fair and controlled fashion among all ABR sources. Any capacity that ABR sources do not use remains available for UBR traffic.
Once source
102
has received permission, it begins scheduling cells for transmission at the allowed cell rate (ACR). The ACR is initially set to the initial cell rate (ICR) and is always bounded by the MCR and the PCR. Transmission of data cells
108
is preceded by transmission of a forward RM cell
110
. Source
102
continues to send forward RM cells
11

Affiliated with

Also associated with

Rating

Explicit rate computation for flow control in compute networks does not yet have a rating. At this time, there are no reviews or comments for this patent.

If you have personal experience with Explicit rate computation for flow control in compute networks, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Explicit rate computation for flow control in compute networks will most certainly appreciate the feedback.

Rate now

Comments { 0 }

Profile ID: LFUS-PAI-O-3019859