Mills, D., "Internet Delay Experiments", RFC-889, M/A-COM Linkabit, December 1983.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....ted in equation 3 is changed only at the start of a new talkspurt. NeVot 1.4 uses a 998002 and b 4 0. This algorithm will henceforth be referred to as Algorithm 1. Certain modifications have been suggested to this algorithm for improving its performance. One modification, due to Mills [8], involves two different values of a one for increasing trends and one for decreasing trends. The playout delay and variation are estimated as per equations 1 and 2. Another modification, from NeVot 1.6) is to use the minimum of all network delays in the previous talkspurt as an estimate of the ....

D. Mills, "Internet delay experiments," ARPANET Working Group Request For Comment, (Decenber 1983), RFC 889.

....delay is greater than the actual delay with a high probability. The ted in equation 3 is changed only at the start of a new talkspurt. NeVot 1.4 uses a 0.998002 and 4.0. This algorithm will henceforth be referred to as Algorithm 1. One modification to the basic algorithm, due to Mills [6], involves two different values of a one for increasing trends and one for decreasing trends. Another modification, from NeVot 1.6) is to use the minimum of all network delays in the previous talkspurt as an estimate of the network delays. That is, di = mines (n) 4) where i is the set of all ....

D. Mills, "Internet delay experiments," ARPANET Working Group Request For Comment, (Decenber 1983), RFC 889.

....retransmit timer computation in RFC2988 brings back the original values. It puts a 1 second minimum on the RTO, and speci es the minimum value of the upper limit to be 60 seconds. 27 Further experience with the Internet showed that the RTO computation above su ered from signi cant shortcomings [109, 110, 127, 189]. A study of Internet delays showed that delay measurements are generally Poisson distributed, except for bursts of delays that are several times larger than typical [127] The estimation algorithm of RFC793 was shown to react too slowly to such bursts, resulting in many false timeouts. A proposed ....

.... Further experience with the Internet showed that the RTO computation above su ered from signi cant shortcomings [109, 110, 127, 189] A study of Internet delays showed that delay measurements are generally Poisson distributed, except for bursts of delays that are several times larger than typical [127]. The estimation algorithm of RFC793 was shown to react too slowly to such bursts, resulting in many false timeouts. A proposed x was to have a large weight applied to samples that are larger than the current SRTT. In addition, the weight used for samples smaller than the SRTT would be small to ....

[Article contains additional citation context not shown here]

Mills D. L., Internet Delay Experiments, RFC889, December 1983.

....the network delays, such as router queue delay and link propagation delay as well as host processing delay, such as the time that takes the sender and receiver to process the packet and the acknowledgment. Normally, the propagation delay is a significant contributor to the round trip time. RFC 889 [11] states that RTTs over the wide area networks with wired links show Pois son distribution characteristics. We conducted experiments on non ambiguous mea surement of RTT values of communication sessions 15 1068 9605 01 0100 0015519.50 0 2001 Ptenum Pubfishing Corporation via both wired and ....

D. L. Mills, Internet delay experiments, Request for Comments (RFC) 889, December 1983.

....a set of (time stamped) messages that yield an estimated (offset, delay) tuple, and (b) by synchronizing clock offsets based on a minimum delay filter. Mills shows that this leads to the minimum error in clock offsets. Mills has also reported on Internet Delays as a function of packet size etc. [23]. Goal: Timer Adjustment: Several papers have addressed timer adjustment in TCP [16, 19, 39] The first two papers model round trip delays as a first order Auto Regressive Moving Average process. The third paper shows some of the difficulties of timer adjustment in TCP. Goal: Report On NSFNet ....

Mills, D.L., "Internet delay experiments," Request for Comment: RFC-889, Net- work Information Center, SRI International, December 1983.

....must be reconfigured in this situation. 11 Future Work It is also possible to analyze the distribution of RTTs (round trip times) to the clients in order to be able to determine a more correct aggressive timeout period. RTTs can be modeled more easily since they are reported to be roughly Poisson[15]. Furthermore, in this case we will be able to change the aggressive timeout period regarding the changing behaviour of RTTs. The possible response of an attacker against this improvement can be training the network monitor with small RTTs (if possible) in order to cause denial of service to ....

D. Mills, Internet Delay Experiments, RFC 889 (1983).

.... over the course of a connection [Nag84, DDK 90] The early TCP specification included a notion of dynamically estimating RTO, based on maintaining an exponentially weighted moving average (EWMA) of the current RTT and a static variation term [Pos81] This estimator was studied by Mills in [Mil83] which characterizes measured Internet RTTs as resembling a Poisson distribution overall, but with occasional spikes of much higher RTTs, and suggests changing the estimator so that it more rapidly adapts to increasing RTTs and more slowly to decreasing RTTs. To our knowledge, this modified ....

....a Poisson distribution overall, but with occasional spikes of much higher RTTs, and suggests changing the estimator so that it more rapidly adapts to increasing RTTs and more slowly to decreasing RTTs. To our knowledge, this modified estimator has not been further evaluated in the literature. Mil83] also noted that the balance between responding rapidly in the face of true loss versus avoiding unnecessary retransmissions appears to be a fundamental tradeoff, with no obvious optimal solution. Zhang [Zha86] discusses a number of deficiencies with the standard TCP RTO estimator: ambiguities in ....

[Article contains additional citation context not shown here]

David Mills. Internet Delay Experiments, Diciembre 1983. RFC 889.

.... over the course of a connection [Nag84, DDK 90] The early TCP specification included a notion of dynamically estimating RTO, based on maintaining an exponentially weighted moving average (EWMA) of the current RTT and a static variation term [Pos81] This estimator was studied by Mills in [Mil83] which characterizes measured Internet RTTs as resembling a Poisson distribution overall, but with occasional spikes of much higher RTTs, and suggests changing the estimator so that it more rapidly adapts to increasing RTTs and more slowly to decreasing RTTs. To our knowledge, this modified ....

....a Poisson distribution overall, but with occasional spikes of much higher RTTs, and suggests changing the estimator so that it more rapidly adapts to increasing RTTs and more slowly to decreasing RTTs. To our knowledge, this modified estimator has not been further evaluated in the literature. Mil83] also noted that the balance between responding rapidly in the face of true loss versus avoiding unnecessary retransmissions appears to be a fundamental tradeoff, with no obvious optimal solution. Zhang [Zha86] discusses a number of deficiencies with the standard TCP RTO estimator: ambiguities in ....

[Article contains additional citation context not shown here]

David Mills. Internet Delay Experiments, December 1983. RFC 889.

....is Retransmission Time Out (RTO) In simple words, RTO is the estimate on the part of the sender on how much time it should take for a segment to be acknowledged. TCP uses adaptive retransmission algorithm to arrive at a reasonable value for RTO. This algorithm is basically an estimation process[Mil83]. TCP maintains a current estimate of RTT. To collect the samples of RTT, TCP records the time at which each segment is sent and the time at which the ACK for that particular segment is received. Subtracting the two values, TCP obtains the current RTT estimate 3 . 3 The exact implementation ....

D.L. Mills. Internet Delay Experiments, 1983. RFC 889.

....studies analyze properties of a significant fraction of the entire network. Examples are Kleinrock s study of the ARPANET s behavior on time scales of hours to days [Kl76] the series of ping experiments conducted by Mills to evaluate the effectiveness of the TCP retransmission timeout algorithm [Mi83]; Claffy et al. s study characterizing traffic on the T1 NSFNET backbone [CPB93b] and Chinoy s study of the dynamics of routing information within the NSFNET backbone [Ch93] While these studies can convincingly characterize the full range of behavior one might expect to observe from the ....

....are no longer the routers relevant to the realtime path. 5. If routes change frequently, then network measurements face difficult consistency problems. For example, several studies of end to end network behavior rely on repeated measurements of a network path made over the course of hours to days [Mi83, CPB93a, Bo93, SAGJ93, Mu94, BCG95]. Whether these measurements all observe the same path significantly affects the accuracy of the studies. Similarly, distributed algorithms for analyzing the network s state also face consistency problems if routes change frequently. For example, recent theoretical work has developed tomography ....

[Article contains additional citation context not shown here]

D. Mills, "Internet Delay Experiments," RFC 889, Network Information Center, SRI International, Menlo Park, CA, 1983.

....a message and using it to determine the typical one way propagation delay. In a local area this solution can produce a very accurate 13 result if knowledge of message processing times is reliable. Unexpected delays can be minimised by providing kernel support for message delay measurement as in [Mills83] or hardware support for time stamping message transmission and arrival [Kopetz87, Damm89] Outside a local area it is much less accurate, because the probability of having a symmetric delay distribution is significantly reduced. Use of a suitable jitter bound in an ATM network provides a ....

D. L. Mills. Internet Delay Experiments. Internet RFC-889, December 1983. (p 14)

.... over the course of a connection [Nag84, DDK 90] The early TCP specification included a notion of dynamically estimating RTO, based on maintaining an exponentially weighted moving average (EWMA) of the current RTT and a static variation term [Pos81] This estimator was studied by Mills in [Mil83] which characterizes measured Internet RTTs as resembling a Poisson distribution overall, but with occasional spikes of much higher RTTs, and suggests changing the estimator so that it more rapidly adapts to increasing RTTs and more slowly to decreasing RTTs. To our knowledge, this modified ....

....a Poisson distribution overall, but with occasional spikes of much higher RTTs, and suggests changing the estimator so that it more rapidly adapts to increasing RTTs and more slowly to decreasing RTTs. To our knowledge, this modified estimator has not been further evaluated in the literature. Mil83] also noted that the balance between responding rapidly in the face of true loss versus avoiding unnecessary retransmissions appears to be a fundamental tradeoff, with no obvious optimal solution. Zhang [Zha86] discusses a number of deficiencies with the standard TCP RTO estimator: ambiguities in ....

David Mills. Internet Delay Experiments, December 1983. RFC 889.

....ff. In our experimental studies, the value of ff was chosen to be 0.998002 to conform with the existing implementation of this algorithm in the distribution of NeVoT 1.4 [13] 3. 2 Algorithm 2 The second algorithm is a small modification to the first algorithm, based on a suggestion by Mills [8] Time of arrival (in seconds) 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.4 0 0.2 0.4 0.6 0.8 1 1.2 Figure 5: A typical spike which would allow the TCP retransmission timer estimate to adapt more quickly to short burst of packets incurring long delays. The idea is to use a different weighting mechanism by ....

....4 In the course of our work, we examined a series of traces of the observed network delay of audio packets transmitted point to point between several Internet sites. In these traces, we noted frequent occurrences of delay spikes. The occurrence of spikes have been reported by Mills and Bolot[8, 2]. Figure 5, adapted from one of these traces, depicts a typical spike. Each point shown represents a packet arriving at the time indicated by its x axis value, having experienced an end to end network delay (n i ) equal to the y axis value. A spike constitutes a sudden, large increase in the ....

D. Mills, "Internet Delay experiments," ARPANET Working Group Requests for Comment, (December 1983), RFC 889.

....of day and days of the week, etc. Other measurements were taken to determine how delays across the ARPANET were influenced by packet length. The results were used to assess whether TCP performance could be improved by including a dependence on packet length in the retransmission timeout algorithm [15]. Several other studies have addressed timeout adjustment in TCP, and they have proposed improvements to take into account packet losses, packet retransmissions, and the variance of packet round trip delays [12, 13] The NSFNET replaced the ARPANET in 1990. Recent studies have measured the delay ....

D. Mills, "Internet delay experiments", RFC 889, December 1983.

....measure (and predict) that time, even a 4 second clock will not defeat this attack. How accurately can the trip time be measured If we assume that stability is good, we can probably bound it within 10 milliseconds or so. Clearly, the Internet does not exhibit such stability over the long term [9] , but it is often good enough over the short term. 2 There is thus an uncertainty of 2500 in the possible value for ISN S . If each trial takes 5 seconds, to allow time to re measure the round trip time, an intruder would have a reasonable likelihood of succeeding in 7500 seconds, and a ....

Mills, D.L. Internet Delay Experiments , RFC 889, 1983. - 16 -

No context found.

Mills, D., "Internet Delay Experiments", RFC-889, M/A-COM Linkabit, December 1983.

No context found.

D. Mills, Internet Delay Experiments, RFC 889 (1983).

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC