Traffic Management for the Available Bit Rate (ABR) Service in ...
Traffic Management for the Available Bit Rate (ABR) Service in ... Traffic Management for the Available Bit Rate (ABR) Service in ...
This approach is aimed to keep the mean constant. But, it not only has the side- e ect of inducing a pdf impulse at zero, but also changes the shape of the pdf, thus increasing the probability of small positive values. The fourth and nal technique is to simply ignore negative values and values greater than the maximum. This approach keeps the shape of the positive part of the pdf intact while not introducing a pdf impulse at zero. If the number of negative values is small, the mean and variance of the distribution would not have changed appreciably. Further, it can be shown that the new distribution is still long-range dependent. We choose the fourth approach (of ignoring negative values and values greater than the maximum) in our simulations. 8.18 Simulation Con guration and Parameters We use the n Source + VBR con guration described in section 8.7 earlier in this chapter. Recall that the con guration has a single bottleneck link shared by the N ABR sources and a VBR VC carrying the multiplexed stream. Each ABR source is a large (in nite) le transfer application using TCP. All tra c is unidirectional. All links run at 149.76 Mbps. The links traversed by the connections are symmetric i.e., each link on the path has the same length for all the VCs. In our simulations, N is 15 and the link lengths are 1000 km in WAN simulations. In satellite simulations, the feedback delay may be 550 ms (corresponds to a bottleneck after the satellite link) or 10 ms (corresponds to a bottleneck before the satellite link). This is illustrated in Figures 8.17 and 8.18. 321
For the video sources, we choose means and standard deviations of video sources to have three sets of values (7.5 Mbps, 7 Mbps), (10 Mbps, 5 Mbps) and (5 Mbps, 5 Mbps). This choice ensures that the variance in all cases is high, but the mean varies and hence the total VBR load varies. The number of video sources (N) is 9 which means that the maximum VBR load is 80% of 149.76 Mbps link capacity. As discussed later the e ective mean and variance (after bounding the generated value to within 0 and 15 Mbps) may be slightly di erent and it a ects the e ciency measure. We also compare certain results with those obtained using an ON-OFF VBR model described in section 8.16. The Hurst parameter which determines the degree of long-range dependence for each video stream is chosen as 0.8 [13]. Recall that when TCP data is encapsulated over ATM, a set of headers and trailers are added to every TCP segment. We have 20 bytes of TCP header, 20 bytes of IP header, 8 bytes for the RFC1577 LLC/SNAP encapsulation, and 8 bytes of AAL5 information, a total of 56 bytes. Hence, every MSS of 512 bytes becomes 568 bytes of payload for transmission over ATM. This payload with padding requires 12 ATM cells of 48 data bytes each. The maximum throughput of TCP over raw ATM is (512 bytes/(12 cells 53 bytes/cell)) = 80.5%. Further in ABR, we send FRM cells once every Nrm (32) cells. Hence, the maximum throughput is 31/32 0.805 = 78% of ABR capacity. For example, when the ABR capacity is 149.76 Mbps, the maximum TCP payload rate is 116.3 Mbps. Similarly, fora MSS of 9140 bytes, the maximum throughput is 87% of ABR capacity. We use a metric called \e ciency" which is de ned as the ratio of the TCP throughput achieved to the maximum throughput possible. As de ned above the 322
- Page 297 and 298: the performance is fair. Also, the
- Page 299 and 300: Figure 8.7(b) shows the rates (ACRs
- Page 301 and 302: Window Size in bytes vfive-tcp/opti
- Page 303 and 304: The e ect of large bu ers on CLR is
- Page 305 and 306: 8.13 Summary of TCP/IP performance
- Page 307 and 308: Feedback delay: Twice the delay fro
- Page 309 and 310: on a link, two cells are expected a
- Page 311 and 312: state only after the switch algorit
- Page 313 and 314: queue length is less susceptible to
- Page 315 and 316: to equalize rates for fairness, and
- Page 317 and 318: QB = Link bandwidth (RT T ; T ) and
- Page 319 and 320: u ered at the end-system, and not i
- Page 321 and 322: Part b): When ABR load goes away, t
- Page 323 and 324: All our simulations presented use t
- Page 325 and 326: Averaging RTT(ms) Feedback Max Q Th
- Page 327 and 328: a modi ed version of the ERICA algo
- Page 329 and 330: ABR is better than UBR in these (en
- Page 331 and 332: problems. During the ON time, the V
- Page 333 and 334: hand, the frequency of the VBR is h
- Page 335 and 336: like ERICA+ which uses the queueing
- Page 337 and 338: minimum fairshare is low. This may
- Page 339 and 340: 8.17 E ect of Long-Range Dependent
- Page 341 and 342: terminology) are called \Presentati
- Page 343 and 344: The key point is that the MPEG-2 ra
- Page 345 and 346: the MPEG-2 Transport Stream, and st
- Page 347: 8.17.4 Observations on the Long-Ran
- Page 351 and 352: Video Sources ABR Metrics # Mean St
- Page 353 and 354: However, with modi cations to ERICA
- Page 355 and 356: Video Sources ABR Metrics # Avg Src
- Page 357 and 358: Video Sources ABR Metrics # Avg Src
- Page 359 and 360: On the other hand, if the applicati
- Page 361 and 362: of managing bu ers, queueing, sched
- Page 363 and 364: hence control the total load on the
- Page 365 and 366: call such a switch a \VS/VD switch"
- Page 367 and 368: Figure 9.2: Per-class queues in a n
- Page 369 and 370: which arises is where the rate calc
- Page 371 and 372: 9.2 The ERICA Switch Scheme: Renota
- Page 373 and 374: The unknowns in the above equations
- Page 375 and 376: Figure 9.9: Two methods to measure
- Page 377 and 378: 9.4 VS/VD Switch Design Options 9.4
- Page 379 and 380: # VC Rate VC Input Rate Input Rate
- Page 381 and 382: 8 uses source rate measurement, we
- Page 383 and 384: The allocated rate update and the e
- Page 385 and 386: sources in chapter 6. We expect the
- Page 387 and 388: con guration mentioned in the table
- Page 389 and 390: can be very di erent for di erent V
- Page 391 and 392: CHAPTER 10 IMPLEMENTATION ISSUES At
- Page 393 and 394: With an enhanced UBR service, appli
- Page 395 and 396: 2. Some switch schemes have a proce
- Page 397 and 398: 4. Large legacy switches have a pro
For <strong>the</strong> video sources, we choose means and standard deviations of video sources<br />
to have three sets of values (7.5 Mbps, 7 Mbps), (10 Mbps, 5 Mbps) and (5 Mbps,<br />
5 Mbps). This choice ensures that <strong>the</strong> variance <strong>in</strong> all cases is high, but <strong>the</strong> mean<br />
varies and hence <strong>the</strong> total VBR load varies. The number of video sources (N) is 9<br />
which means that <strong>the</strong> maximum VBR load is 80% of 149.76 Mbps l<strong>in</strong>k capacity. As<br />
discussed later <strong>the</strong> e ective mean and variance (after bound<strong>in</strong>g <strong>the</strong> generated value to<br />
with<strong>in</strong> 0 and 15 Mbps) may be slightly di erent and it a ects <strong>the</strong> e ciency measure.<br />
We also compare certa<strong>in</strong> results with those obta<strong>in</strong>ed us<strong>in</strong>g an ON-OFF VBR model<br />
described <strong>in</strong> section 8.16.<br />
The Hurst parameter which determ<strong>in</strong>es <strong>the</strong> degree of long-range dependence <strong>for</strong><br />
each video stream is chosen as 0.8 [13].<br />
Recall that when TCP data is encapsulated over ATM, a set of headers and trailers<br />
are added to every TCP segment. We have 20 bytes of TCP header, 20 bytes of IP<br />
header, 8 bytes <strong>for</strong> <strong>the</strong> RFC1577 LLC/SNAP encapsulation, and 8 bytes of AAL5<br />
<strong>in</strong><strong>for</strong>mation, a total of 56 bytes. Hence, every MSS of 512 bytes becomes 568 bytes<br />
of payload <strong>for</strong> transmission over ATM. This payload with padd<strong>in</strong>g requires 12 ATM<br />
cells of 48 data bytes each. The maximum throughput of TCP over raw ATM is (512<br />
bytes/(12 cells 53 bytes/cell)) = 80.5%. Fur<strong>the</strong>r <strong>in</strong> <strong>ABR</strong>, we send FRM cells once<br />
every Nrm (32) cells. Hence, <strong>the</strong> maximum throughput is 31/32 0.805 = 78% of<br />
<strong>ABR</strong> capacity. For example, when <strong>the</strong> <strong>ABR</strong> capacity is 149.76 Mbps, <strong>the</strong> maximum<br />
TCP payload rate is 116.3 Mbps. Similarly, <strong>for</strong>a MSS of 9140 bytes, <strong>the</strong> maximum<br />
throughput is 87% of <strong>ABR</strong> capacity.<br />
We use a metric called \e ciency" which is de ned as <strong>the</strong> ratio of <strong>the</strong> TCP<br />
throughput achieved to <strong>the</strong> maximum throughput possible. As de ned above <strong>the</strong><br />
322