Early Packet Discard Research Articles

On-demand packet discard scheme for TCP over ATM-UBR service

The performance of the TCP over ATM networks is investigated. A TCP packet is segmented into multiple ATM cells at the sending node and reassembled at the receiving node. When congestion occurs, the ATM switch may drop some of the cells. If one or more cells of a TCP packet are dropped, the TCP packet is said to be corrupted. A corrupted packet will be discarded at the receiving node. Transmitting the cells of a corrupted packet will only waste network resources and lower the effective throughput. A major objective of a selective packet discard scheme is to minimise the packet fragmentation problem. Previous approaches such as the early packet discard (EPD) scheme and its variants may drop incoming cells when congestion is anticipated. As a result, the buffer is under-utilised most of the time. Packets may be dropped unnecessarily, and the fragmentation problem cannot be avoided. A new packet discard scheme, called the on-demand packet discard (ODPD) scheme, is presented by the author. The ODPD scheme can fully utilise the available buffer space and effectively avoid the fragmentation problem. The performance of ODPD is found to be better than EPD and its variants under all traffic conditions.

A study of a new selective cell discard scheme with two threshold control on ATM networks

AbstractWhen UBR (Unspecified Bit Rate) service is used on an ATM network, cells are discarded in congestion and imperfect packets are generated that consume network resources in a wasteful manner. As a remedy, the EPD (Early Packet Discard) control is proposed, in which the cells are discarded on a packet basis in congestion by providing a threshold to the input buffer. However, with EDP control, there is a possibility that the EOP (End Of Packet) cell in AAL5 may be discarded. Thus, two successive packets become invalid. For this reason, the TT‐EPD (Two Threshold‐EPD) that prevents discarding of the EOP cell has been proposed. This scheme sets two threshold values (TH1 and TH2) and the fixed TH1 works similarly to EPD control. TH2 carries out dynamic control and secures the input region for the EOP cell in the buffer. In this scheme, the amount of buffer between TH1 and TH2 decreases and there is an increase in the number of acceptance failures in which the cells composing a packet initially accepted are discarded by TH2. In this paper, a method that keeps the amount of buffer between TH1 and TH2 constant to reduce the acceptance missing packets is proposed. The effectiveness of the proposed scheme is demonstrated by simulation. © 2003 Wiley Periodicals, Inc. Electron Comm Jpn Pt 1, 86(8): 50– 57, 2003; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/ecja.10039

SWFA: a new buffer management mechanism for TCP over ATM-GFR

We investigate packet discarding schemes for transport control protocol (TCP) over asynchronous transfer mode with guaranteed frame rate service. In this letter, we propose the selective weighted fair allocation (SWFA) scheme, which discards packets from selected sessions. 15 TCP connections with equal minimum cell rate (MCR) and unequal MCRs are simulated. The SWFA scheme with per-virtual connection (VC) queuing is compared with early packet discard (EPD) with first-in, first-out queuing, EPD with per-VC queuing, and differential fair buffer allocation with per-VC queuing. Our results show that SWFA with per-VC queuing achieves significant enhancement on throughputs, goodputs, and fairness.

Fair scheduling in Internet routers

Input buffered switch architecture has become attractive for implementing high performance routers and expanding use of the Internet sees an increasing need for quality of service. It is challenging to provide a scheduling technique that is both highly efficient and fair in resource allocation. In this paper, we first introduce an iterative fair scheduling (IFS) scheme for input buffered switches that supports fair bandwidth distribution among the flows and achieves asymptotically 100 percent throughput. The IFS is evaluated both under synthetic workload and with Web traces from the Internet. Compared to the commonly used synthetic input, our simulation results reveal significant difference in performance when the real network traffic is employed. We then consider fair scheduling under various buffer management mechanisms and analyze their impact on the fairness in bandwidth allocation. Our studies indicate that early packet discard in anticipation of congestion is necessary and per-flow based buffering is effective for protecting benign users from being adversely affected by misbehaving traffic. Buffer allocation according to bandwidth reservation is especially helpful when the input traffic is highly bursty.

ATM기반 MPLS망에서 VC-Merge 가능한 고속 스위치 구현에 관한 연구

본 논문에서는 ATM 기반 MPLS(Multi-Protocol Label Switch)망에서 라우터의 레이블 공간을 효율적으로 사용하여 망의 확장성을 높이기 위한 방안인 레이블 통합 기능이 탑재된 고속 Crossbar Switch론 구현한다. 차등서비스(Differentiated Service)와 레이블 통합 기능을 동시에 수행 할 수 있는 적합한 하드웨어 구조를 제안한다. 본 논문에서는 각 코어 LSR(Label Switch Router)의 출력버퍼에서 망 폭주 발생 가능성이 있을 시 EPD(Early Packet Discard) 알고리즘을 통한 적응적 폭주 제어 방법을 사용하므로써 네트워크 자원의 낭비론 막았으며, 제안한 VC(Virtual Channel)-merge 기법의 정당성을 입증하기 위하여 Non VC-merge 기법과의 비교 분석을 시뮬레이션을 통하여 수행하였다. 제안한 VC-merge가능한 스위치는 VHDL로 모델링하여 합성 설계하고, 삼성 0.5m SOG 공정으로 팁을 제작하였다. In this paper, we implement a high-speed swatch tilth the function for label integration to enhance the expansion of networks using the label space of routers efficiently on MPLS over ATM networks. We propose an appropriate hardware structure to support the VC-merge function and differentiated services simultaneously. In this paper, we use the adaptive congestion control method such as EPD algorithm in carte that there is a possibility of network congestion in output buffers of each core LSR. In addition, we justify the validity of the proposed VC-merge method through simulation and comparison to conventional Non VC-merge methods. The proposed VC-merge capable switch is modeled in VHDL. synthesized, and fabricated using the SAMSUNG 0.5um SOG process.

Performance Analysis of Generalized Selective Packet Discarding Schemes

Selective packet discarding schemes have been mostly developed and analyzed in the context of ATM networks. However, similar schemes could be applied to non-ATM or more general networks. The analysis of selective packet discarding schemes in a general packet switching environment has been limited to the basic dropping schemes, i.e., Partial Packet Discard (PPD) and Early Packet Discard (EPD), and for only the single traffic source case. In this paper, we extend the mathematical analysis of PPD and EPD for multiple sources and include simulation results to verify the analytical results. Finally, to place the performance of the generalized EPD and PPD in the context of other packet dropping schemes, we also provide simulation results of the Drop from Front (DfF), and preemptive Partial Packet Discard (pPPD) schemes.

A performance model of partial packet discard and early packet discard schemes in ATM switches

In this paper, we develop a concise performance model of partial packet discard (PPD) and early packet discard (EPD) schemes in ATM switches. We study the performance of PPD and EPD with heterogeneous traffic sources. The sources included Poisson, and ON–OFF with long-tailed sojourn time distribution, which is approximated by a hyperexponential distribution. The fairness of EPD is investigated. We automatically generate and numerically solve the underlying Markov chain using a high-level graphical paradigm known as the stochastic reward net. Our numerical results reveal that: (1) the benefit of PPD and EPD is not significant when the queuing system is underloaded with Poisson sources; (2) PPD and EPD can increase the goodput when the system is overloaded or loaded with ON–OFF sources; (3) in All Poisson case, PPD and EPD provide nearly fair service to the sources; (4) the ON–OFF source gets higher goodput than the Poisson source. Because the burstiness of a source will be alleviated by the statistical multiplexing of ATM switches, the ON–OFF source may be viewed as the source that just enters the network, and the Poisson source may be viewed as the source being far away from the network node we are considering and has been regulated and smoothed by the switches it traversed. Therefore, the queuing system with EPD algorithm gives a higher goodput to the sources near it. And the sources far away from the system will have lower goodput. One of our principal conclusions is that per-VC-based scheme is not needed at the core of large ATM networks because the traffic is less bursty at the core of the networks and PPD and EPD are fair under this kind of environment. A per-VC-based scheme may be used at the edge of the ATM cloud. Such a configuration can make the core of the ATM network work at high speed.

Balanced packet discard for improving TCP performance in ATM networks

TCP suffers from low performance over asynchronous transfer mode (ATM) networks. This is mainly because during phases of congestion, ATM drops cells without taking into account the effect this has on the upper layer protocols. Two main algorithms, called partial packet discard (PPD) and early packet discard (EPD), were proposed in the past for improving TCP performance. However, they address one aspect of the problem that has only small effect on the final performance. In this paper we propose an enhanced method for packet discard, called balanced packet discard (BPD) that improves TCP performance dramatically on congested networks and guarantees fairness among multiple connections. We will show that BPD increases TCP throughput by more than 25% compared to EPD/PPD.

Buffer management and scheduling schemes for TCP/IP over ATM‐GFR

AbstractToday ATM technology is facing challenges from integrated service IP, IP switching, gigabit IP router and gigabit ethernet. Although ATM is approved by ITU‐T as the standard technology in B‐ISDN, its survivability is still in question. Since ATM‐UBR (unspecified bit rate) provides no service guarantee and ATM‐ABR (available bit rate) is still unattainable for most users, many existing users have little or no incentives to migrate to ATM technology. The guaranteed frame rate (GFR) service is introduced to deal with this dilemma. The GFR can guarantee the minimum cell rate (MCR) with fair access to excess bandwidth. This paper studies various schemes to support the GFR. We have studied different discarding and scheduling schemes, and compared their throughput and fairness when TCP/IP traffic is carried. Through simulations, it is shown that only per‐VC queueing with weighted round robin (WRR) can guarantee minimum cell rate. Among all the schemes that have been explored, we recommend dynamic threshold–early packet discard (DT–EPD) integrated with MCR+ (a WRR variant) to support the GFR service. Copyright © 2001 John Wiley & Sons, Ltd.

Improving goodputs of IP packets under ATM UBR traffic with port-based queueing schemes

ATM networks are widely adopted as backbone networks. TCP/IP are the most popular protocols today and it is quite possible that packets are forwarded by ATM networks. In ATM networks, UBR (unspecified bit rate) service and ABR (available bit rate) service are common in transmitting IP packets. Because TCP has its own flow control mechanism, the flow control of ABR service could become redundant. Moreover, it could interfere with the TCP flow control. In this paper, we will focus on buffer management schemes of ATM switches with UBR service. Researchers have proposed several buffer management schemes for UBR service. However, all schemes do not consider the influence of output rates. In addition, most researchers assume that each port has its own buffers. A basic scheme called dedicated buffers per port is compared with four proposed schemes. The paper considers the situation that switch buffers are shared among all output ports and proposes a scheme called shared buffers among all ports. Then, two schemes, dedicated buffers with early packet discarding (EPD) and shared buffers with EPD, are proposed. The EPD scheme is modified. Instead of using a threshold for each flow, only one threshold is applied to buffers of each output port. The fifth proposed scheme is called shared buffers with EPD + differential buffer allocation (DBA), which also considers the effects of various output rates. Results show that goodputs are increased apparently by applying the DBA scheme. Results also show that for goodputs, when packets pass two or more switches, packet discarding schemes are far more important than whether we adopt schemes of shared buffers or schemes of dedicated buffers. Hence, we can simply adopt schemes of dedicated buffers, which are easy to implement, and incorporate a packet-discarding scheme in most cases.

Performance of TCP applications over ATM networks with ABR and UBR services—a simulation analysis

This paper aims to present a comparative simulation study of the performance of TCP traffic over ATM networks with UBR and ABR services; to identify weaknesses of currently most promising ATM congestion control schemes; and to indicate the requirements for a fairer, simpler and more robust congestion control mechanism. Variety of congestion control schemes under various network configurations and traffic patterns are simulated for a comprehensive analysis. Simulation results show that Early Packet Discard (EPD) wastes buffer capacity and has no provision for fairness, Intelligent Marking [K.Y. Siu, H.Y. Tzeng, Computer Communication Review 24 (5) (1995) 81–106] does not provide adequate fairness among competitive connections, and ERICA+ [R. Jain, S. Kalyanaraman, Y. Goyal, S. Fahmy, R. Viswanathan, ERICA switch algorithm: a complete description, ATM Forum Contribution 96-1172, 1996] is too restrictive and unnecessarily complicated. The paper also indicates that TCP flow control and ATM congestion control do not cooperate well when buffer overflows and retransmission occurs, necessitating further control mechanisms for improving performance.

Excess buffer requirement for EPD schemes in ATM networks

It is known that the performance of TCP over ATM can be significantly degraded if the bandwidth is occupied by cells belonging to packets that are already corrupted by cell loss due to buffer overflow. The Early Packet Discard (EPD) mechanism is a well known technique to improve the performance of TCP or other transport layer protocols over ATM by allowing only non-corrupted packets to be transmitted. The basic idea of EPD is to discard an entire packet prior to buffer overflows; a switch will drop incoming packets whenever the buffer occupancy exceeds a pre-defined EPD threshold. The excess buffer capacity for the EPD mechanism refers to the available buffer capacity above the EPD threshold in a switch. The excess buffer is needed to accommodate the outstanding cells that are not discarded. Moreover, sufficient excess buffer is important in maintaining good throughput performance. In this article, we shall present analytical results on the worst-case excess buffer capacity for EPD mechanism that incorporate per-VC accounting and per-VC queueing techniques, extending the work of Turner (J.-S. Turner, Maintaining High Throughput During Overloading In ATM Switches, IEEEINFOCOM’96, March 1996) for the plain EPD case. We derive tight upper bounds on the excess buffer occupancy in per-VC based EPD schemes. Our analytical results show that per-VC based EPD schemes can provide better performance at the expense of more excess buffer than the plain EPD scheme. Simulation results are also presented and shown to be in agreement with the theoretical results. We also provide a simulation study on more general cases for both plain EPD and per-VC based EPD schemes, and reasonable rules of thumb for choosing excess buffer in these situations are discussed.

A RED discard strategy for ATM networks and its performance evaluation with TCP/IP traffic

In ATM UBR networks supporting TCP traffic, optimal efficiency can only be envisaged if switches adopt a discard mechanism that operates at the packet level rather than the cell level. In this paper, we define a variant of the RED discard strategy suitable for ATM switches. An interesting feature of this ATM-RED is that it has a similar per-VC implementation complexity as the Early Packet Discard (EPD) algorithm. To study the efficiency of the ATM-RED discard strategy, we compare its performance with plain the UBR, EPD and Fair Buffer Acceptance (FBA) discard strategies by means of simulation with TCP/IP traffic. We give comparative results with respect to different performance criteria such as goodput and fairness in various environments, such as end-to-end ATM networks and IP-based networks with an ATM backbone, in both single-bottlenecked and GFC topologies.

Performance analysis of data packet discarding in ATM networks

Data performance in ATM networks should be measured on the packet level instead of the cell level, since one or more cell losses within each packet is equivalent to the loss of the packet itself. Two packet-level control schemes, packet tail discarding and early packet discarding, were proposed to improve data performance. In this paper, a new stochastic modeling technique is developed for performance evaluation of two existing packet-discarding schemes at a single bottleneck node. We assume that the data arrival process is independent of the nodal congestion, which may represent the unspecified bit-rate traffic class in ATM networks, where no end-to-end feedback control mechanism is implemented. Through numerical study, we explore the effects of buffer capacity, control threshold, packet size, source access rate, underlying high-priority real-time traffic, and loading factor on data performance, and discuss their design tradeoffs. Our study shows that a network system can he entirely shut down in an overload period if no packet-discarding control scheme is implemented, under the assumption that there are no higher layer congestion avoidance schemes. Further, unless with sufficiently large buffer capacity, early packet discarding (EPD) always outperforms packet tail discarding (PTD) significantly under most renditions. Especially under the overload condition, EPD can always achieve about 100% goodput and 0% badput, whereas the PTD performance deteriorates rapidly. Among all the factors, the packet size has a dominant impact on EPD performance. The optimal selection of the EPD queue control threshold to achieve the maximum goodput is found to be relatively insensitive to traffic statistics.

Improved WWW multimedia transmission performance in HTTP/TCP over ATM networks

Transmission control protocol/Internet protocol (TCP/IP) is the de facto standard of the networking world. It dynamically adjusts routing of packets to accommodate failures in channels and allows construction of very large networks with little central management. But IP packets are based on the datagram model and are not really suited to real-time traffic. In order to overcome the drawbacks, a new network technology, ATM, is proposed. ATM provides quality of service (QOS) guarantees for various classes of applications and in-order delivery of packets via connection oriented virtual circuits. Unfortunately, when ATM is to be internetworked with the existing network infrastructure, some special signaling, addressing and routing protocols are needed. IP over ATM is one of the methods proposed by IETF. It allows existing TCP/IP applications to run on ATM end-stations and ATM networks to interconnect with legacy LAN/WAN technologies. But the performance of TCP/IP over ATM leaves something to be desired. Partial packet discard (PPD) and early packet discard (EPD) are two schemes to improve its performance. This paper proposes a "selective packet retransmission" scheme for improving HTTP/TCP performance when transmitting through ATM networks. In selective packet retransmission, we take advantage of the property of humans' perception tolerance for errors to determine whether to retransmit a corrupted TCP segment or not. For lossable data, such as images, when an error occurs because of cell losses, it will not be retransmitted. The simulations show that, for the same buffer size and traffic load, selective packet retransmission performs better than PPD, EPD, and plain TCP over ATM.

Performance of TCP traffic and ABR congestion control mechanisms

This paper presents a performance investigation of TCP traffic carried over ATM networks when the available bit rate (ABR) congestion control mechanism is exercised at the ATM layer. Two ABR schemes are investigated; the EFCI marking scheme and the relative rate marking scheme. The joint effect of the ABR mechanism with the early packet discard (EPD) mechanism is also considered. The focus of this study is to evaluate these schemes based on the effective throughput experienced by end users, taking the effect of retransmissions into account. Simulation results are obtained for a multihop switch configuration. Various aspects regarding the synergy of the TCP and ABR flow control mechanisms are presented.

Performance of TCP over UBR in ATM with EPD and virtual queuing techniques

While both available bit rate (ABR) and unspecified bit rate (UBR) services can be used to support data traffic in asynchronous transfer mode (ATM), many ATM switch vendors consider UBR to be more attractive because of its low implementation cost in comparison with ABR. However, since there is no congestion control at the cell level in UBR and cells are simply discarded when buffer overflow occurs, the effective throughput of transmission control protocol (TCP) can degrade significantly over UBR service in a congested network. Early Packet Discard (EPD) techniques have been proposed by Romanow and Floyd (Dynamics of TCP traffic over ATM networks, IEEE Journal on Selected Areas in Communications 13 (4) (1995) 633–641) and shown to improve the throughput of TCP over ATM. In our earlier work (A simulation study of TCP performance in ATM networks with ABR and UBR services, in: Proceedings of the International Phoenix Conference on Computers and Communications, March 1996), we have shown that TCP with EPD can suffer significant degradation in fairness of throughputs among competing VCs in a congested ATM LAN environment, but the degree of fairness can be improved by using per-VC accounting/queuing techniques. In this paper, we continue our studies in designing better and low cost methods for the support of TCP over ATM. We apply the virtual queuing technique proposed in Chiussi et al. (Virtual queuing techniques for ABR service: improving ABR/VBR interaction, Infocom'97) to emulate on a FIFO queue the service provided by per-VC queuing. This technique requires low implementation cost and delivers performance of TCP over UBR comparable to the expensive per-VC queuing technique. Simulation results are presented to demonstrate that our technique combined with EPD can drastically improve the performance of TCP over UBR service.

Improving the performance of TCP over the ATM-UBR service

In this paper we study the design issues for improving TCP performance over the ATM UBR service. ATM-UBR switches respond to congestion by dropping cells when their buffers become full. TCP connections running over UBR can experience low throughput and high unfairness. Intelligent switch drop policies and end-system policies can improve the performance of TCP over UBR with limited buffers. We describe the various design options available to the network as well as to the end systems to improve TCP performance over UBR. We study the effects of early packet discard, and present a per-VC accounting-based buffer management policy. We analyze the performance of the buffer management policies with various TCP end system congestion control policies, including slow start and congestion avoidance, fast retransmit and recovery and selective acknowledgments. We present simulation results for various small and large latency configurations with varying buffer sizes and number of sources.

On the performance of early packet discard

In a previous paper, one of the authors gave a worst case analysis for the early packet discard (EPD) technique for maintaining packet integrity during overload in ATM switches. This analysis showed that to ensure 100% goodput during overload under worst case conditions requires a buffer with enough storage for one maximum length packet from every active virtual circuit. This paper refines that analysis, using assumptions that are closer to what we expect to see in practice, and examines how EPD performs when the buffer is not large enough to achieve 100% goodput. We show that 100% goodput can be achieved with substantially smaller buffers than predicted by the worst case analysis, although the required buffer space can be significant when the link speed is substantially higher than the rate of the individual virtual circuits. We also show that high goodputs can be achieved with more modest buffer sizes, but that EPD exhibits anomalies with respect to buffer capacity, in that there are situations in which increasing the amount of buffering can cause the goodput to decrease. These results are validated by comparison with simulation.

Packet loss performance of selective cell discard schemes in ATM switches

In transport-layer protocols such as TCP over ATM networks, a packet is discarded when one or more cells of that packet are lost, and the destination node then requires its source to retransmit the corrupted packet. Therefore, once one of the cells constituting a packet is lost, its subsequent cells of the corrupted packet waste network resources. Thus, discarding those cells will enable us to efficiently utilize network resources, and will improve the packet loss probability. We focus on tail dropping (TD) and early packet discard (EPD) as selective cell discard schemes which enforce the switches to discard some of the arriving cells instead of relaying them. We exactly analyze the packet loss probability in a system applying these schemes. Their advantages and limits are then discussed based on numerical results derived through the analysis.