TKK | Tietoverkkolaboratorio | Tutkimus
There are ATM nodes from several vendors in the market and some operators even are offering public ATM services. However, in order to really make a breakthrough of ATM we must develop ATM applications which are acceptable also by others than computer-wizards. Our paper gives an outline of the ATM evolution from separate nodes to real and practical services utilizing ATM technology. One of the most prominent experiences of current situation is continuously evolving standards. First adopters have faced a dilemma of never ending software and hardware updates. Although we may except that Internet will be upgraded to ATM because of increasing capacity and quality requirements, the definitive role of ATM in the whole network architecture is still shrouded in mystery, and a lot of practical and theoretical questions are open.
This paper investigates performance evaluation of Asynchronous Transfer Mode (ATM) switching systems: Private Branch Exchanges (PBX), switches and cross-connects. To evaluate performance of a ATM switching systems is not a straightforward as one might expect. This is due to the fact that performance can be seen from different angles. Measurements presented in this paper include: throughput, switching delay, switching delay variation, buffer size, accuracy of the Usage Parameter Control (UPC) procedure and cell discarding under overload situation. These properties have been evaluated through a series of measurements. Measurements are quite general although they have been carried out only against Fore System’s ASX-200 PBX. The main value of this presentation is in comparison of different measurement techniques rather than in the actual results. The results depend on the environment but the techniques are valid in general.
This presentation considers some practical issues of traffic management in ATM (Asynchronous Transfer Mode) networks. Firstly, we illustrate a fundamental dilemma of statistical multiplexing in ATM networks when small loss ratio, small delay and high utili-zation are expected at the same time. Although there are a lot of solutions to this problem, all of them have some disadvantages. Therefore it is necessary to offer several service types meeting the differing user demands. A minimal amount of service types is desirable because of the additional expenses of nodes and network management when implementing several services. The proposal in this paper is to offer only two basic services: the first one covers real-time CBR (Constant Bit Rare) and VBR (Variable Bit Rate) services and the other non-real-time service is aimed especially for the current Internet applications. The main content of this paper is to present the traffic management tools needed for the real implementation of these services.
In this paper, several performance measurements of ATM (Asynchronous Transfer Mode) networks are performed and analysed. Especially, measurement of the buffer size, the traffic management functions and the cell transfer delay in a switching system are studied as a function of various traffic patterns and loading levels. These values have great importance for establishing reliable media for broadband traffic and contributing to the implementation of services in ATM networks.
ATM has been under a thorough standardization process for more than ten years. Looking at it now, what have we achieved during this time period? Originally ATM was meant to be an easy and an efficient protocol enabling varying services over a single network. What it is turning to be is ‘yet another ISDN’-- network full of hopes and promises but too difficult to implement and expensive to market. The fact is that more and more ‘nice features’ are implemented on the cost of overloading network with hard management procedures. Therefore we need to adopt a new approach. This approach keeps a strong reminds on ‘what is necessary.' This paper presents starting points for an alternative approach to the traffic management. We refer to this approach as ‘the minimum management principle.' Choosing of the suitable service classes for the ATM network is made difficult by the fact that the more services one implements the more management he needs. This is especially true for the variable bit rate connections that are usually treated based on the stochastic models. Stochastic model, at its best, can only reveal momentary characteristics in the traffic stream not the long range behavior of it. Our assumption is that ATM will move towards Internet in the sense that strict values for quality make little or no sense in the future. Therefore stochastic modeling of variable bit rate connections seems to be useless. Nevertheless we see that some traffic needs to have strict guarantees and that only economic way of doing so is to use PCR allocation.
One of the most important network elements in the Internet are the routers which do relaying of IP packets. Because of growth of the Internet routers currently experience serious problems in relaying traffic in a satisfying speed. The idea of switching Internet traffic flows has recently been introduced and a new technology called IP switching has emerged. Several differing technological solutions have been suggested. In this paper we describe and compare two methods in flow-based IP switching to make the decisions whether to switch internet traffic flows to separate ATM-connections. Traffic measurements are made in two networks of varying size and based on a specific three-stage flow analysis we suggest that the decision to switch should be made as flexible as possible due to the expected diversity of traffic profiles in different parts of the network. This way the optimal service cluster could be switched and router resources could be optimally utilized. A simple model to determine workload to an IP switch is introduced. Using this model we see that the workload of the flow setup component and the routing component may be optimized, if we use flexible methods to determine the flows that are to be separately switched.
Key issues in the current development of Internet seem to be its capability to scale and to support new real-time or near real-time applications like video- and audio conferencing. There are two factors that affect these qualities: one is the ability to distinguish which connections should be switched and the other is the effective control over network resources. ATM is a serious attempt to standardize global multiservice networks. This attempt seems to suit well for the future Internet. ATM was originally meant to be an easy and an efficient protocol but it is now turning to be ‘yet another ISDN’. More and more features are implemented to ATM resulting in the overloading of the network with management procedures. Therefore a new approach needs to be taken. In this approach a strong reminder of ‘what is necessary’ needs to be kept in mind. This paper presents an alternative, simpler approach to the ATM traffic management and introduces some suggestions how to map Internet applications to simplified ATM environment using an advanced IP switching concept.
Understanding the nature of data communications requires tools to collect data from the computer, the network and their interaction. A tool is needed to get better understanding of the processes generating traffic to the network.
The main components are an instrumented Linux kernel, a synthetic benchmark program, a system call tracer and a set of analysis programs for the post processing. The data collected from the network can be synchronized with the data collected from computer with an adequate accuracy without expensive hardware.
Changes on the operating system (e.g. scheduling algorithm) or on the network can be easily evaluated by the synthetic benchmark where it is possible to modify CPU/IO-intensity ratio and the number of processes each type thus emulating different real-world applications. The data and the code size can be modified to evaluate the memory system performance over different working set sizes.
The early measurements on the Ethernet indicate that this toolbox is useful. Measurements have revealed how network traffic is affected as number of processes changes. The toolbox development continues on ATM environment.
The instrumented kernel patch.
ATM and ATM measurements are rather new area in the telecommunications technology. So far ATM measurements have been apcomblished as a part of product development. This has lead the view to the very narrow area of ATM measurements. This largely due to the fact that there is no clear picture about the ATM traffic and how to model and implement it to the measurement tools. There have been some performance tests between different ATM switching systems but their perspective have been more or less in the area of data communications. This thesis tries to give larger perspective to the problem we are dealing with the traffic management and performance.
From the performance measurements, parameters which have relation to the information transfer are presented and respective ways to measure them. Parameters considered here are transfer delay, usage parameter control and buffering. All of these measurements have been done in manner that they give short insight to the idea.
Keywords: ATM, B-ISDN, QoS, TRAFFIC MEASUREMENT, UPC
A toolbox was developed to get a better understanding of the processes generating traffic to the network. Using an instrumented Linux kernel and a benchmark program network traffic was generated and captured from the network traffic using several combinations of sending and background processes. The traces include network traffic with timestamps, timestamps of system calls from the benchmark program and information about the process and kernel status at the process switch time. By synchronizing and combining the traces from different sources one could understand the interaction between the operating system and the network.
The traffic was characterized by several methods, among other by the usage parameter control used in ATM networks. Linux operating system gives the same service for both CPU and I/O intensive traffic: the bottleneck resource (network) can be unutilized for long periods. In order to make the traffic smoother, one must either add buffering or modify the process scheduling. Several alternatives for scheduling control mechanism were assessed.
Tietoverkkolaboratorio on nyt osa Tietoliikenne- ja tietoverkkotekniikan laitosta. Tällä sivulla oleva tieto voi olla vanhentunutta.
|
Tämän sivun sisällöstä vastaavat ja
Webmaster.
Sivua on viimeksi päivitetty 30.10.1997 15:24. URI: http://www.netlab.tkk.fi/tutkimus/mitta/abstract.shtml [ TKK > Sähkö- ja tietoliikennetekniikan osasto > Tietoverkkolaboratorio > Tutkimus ] |