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 ...
Averaging Averaging Averaging d p(ms) Max # Interval of Na on ? of z on ? Switch (T ms, ( n = 0.9) ( z = 0.2) Queue n cells) (cells) 1. (1,100) YES YES 0.7 20 5223 2. (5,500) YES NO 0.7 20 5637 Table 8.9: E ect of Switch Scheme Row 1 shows the performance with the averaging of Na and z turned on on a for- merly divergent case. Observe that the queue converges and is small. The parameter z is 0.2, which is roughly equivalent to increasing the averaging interval length by a factor of 5. Hence, we try the value (5 ms, 500 cells) as the averaging interval length, without the averaging of overload factor. Row 2 shows that the queue for this case also converges and is small. 8.16.2 Summary of ON-OFF VBR background e ects In this section, we have studied the impact of ON-OFF VBR background tra c on switch bu ering for ABR service carrying TCP tra c. We nd that the ON-OFF times, the feedback delays, and a switchscheme sensitivetovariation in ABR load and capacity may combine to create worst case conditions where the ABR queues diverge. We have motivated three enhancements to the ERICA+ scheme. The modi cations reduce the e ect of the variation and allow the convergence of the ABR queues, without compromising on the e ciency. In the next section, we shall examine the e ect of VBR carrying long-range dependent tra c (similar to multiplexed MPEG-2 tra c) on ABR, and show that the bu er requirements are unchanged. 311
8.17 E ect of Long-Range Dependent (LRD) VBR background tra c We have studied the ABR model extensively with di erent source tra c patterns like persistent sources, ON-OFF bursty sources, ping pong sources, TCP sources and source-bottlenecked VCs. Many of these studies have also considered the performance in the presence of ON-OFF VBR background tra c. In reality, VBR consists of multiplexed compressed audio and video application tra c, each shaped by leaky buckets at their respective Sustained Cell Rate (SCR) and Peak Cell Rate (PCR) parameters. Compressed video has been shown to be long- range dependent by nature [13]. Compressed audio and video streams belonging to a single program are expected to be carried over an ATM network using the MPEG-2 Transport Stream facility asoutlined in reference [34]. In this section, we rst present amodelofmultiplexed MPEG-2 transport streams carried over ATM using the VBR service. Each stream exhibits long-range depen- dence, i.e., correlation over large time scales. We then study the e ect of this VBR background on ABR connections carrying TCP le transfer applications on WAN and satellite con gurations. The e ect of such VBR tra c is that the ABR capacity is highly variant. We nd that a proper switch algorithm like ERICA+ can tolerate this variation in ABR capacity while maintaining high throughput and low delay. We will present simulation results for terrestrial and satellite con gurations. 8.17.1 Overview of MPEG-2 over ATM In this section, we give a short introduction to the MPEG-2 over ATM model and introduce some MPEG-2 terminology. For a detailed discussion, see reference [77]. 312
- Page 287 and 288: 4 make it to the destination are ar
- Page 289 and 290: Figure 8.3: At the ATM layer, the T
- Page 291 and 292: issues and e ect of bursty applicat
- Page 293 and 294: from the loss and they trigger the
- Page 295 and 296: 8.8 Performance Metrics We measure
- 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: minimum fairshare is low. This may
- 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 and 348: 8.17.4 Observations on the Long-Ran
- Page 349 and 350: For the video sources, we choose me
- 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
Averag<strong>in</strong>g Averag<strong>in</strong>g Averag<strong>in</strong>g d p(ms) Max<br />
# Interval of Na on ? of z on ? Switch<br />
(T ms, ( n = 0.9) ( z = 0.2) Queue<br />
n cells) (cells)<br />
1. (1,100) YES YES 0.7 20 5223<br />
2. (5,500) YES NO 0.7 20 5637<br />
Table 8.9: E ect of Switch Scheme<br />
Row 1 shows <strong>the</strong> per<strong>for</strong>mance with <strong>the</strong> averag<strong>in</strong>g of Na and z turned on on a <strong>for</strong>-<br />
merly divergent case. Observe that <strong>the</strong> queue converges and is small. The parameter<br />
z is 0.2, which is roughly equivalent to <strong>in</strong>creas<strong>in</strong>g <strong>the</strong> averag<strong>in</strong>g <strong>in</strong>terval length by a<br />
factor of 5. Hence, we try <strong>the</strong> value (5 ms, 500 cells) as <strong>the</strong> averag<strong>in</strong>g <strong>in</strong>terval length,<br />
without <strong>the</strong> averag<strong>in</strong>g of overload factor. Row 2 shows that <strong>the</strong> queue <strong>for</strong> this case<br />
also converges and is small.<br />
8.16.2 Summary of ON-OFF VBR background e ects<br />
In this section, we have studied <strong>the</strong> impact of ON-OFF VBR background tra c<br />
on switch bu er<strong>in</strong>g <strong>for</strong> <strong>ABR</strong> service carry<strong>in</strong>g TCP tra c. We nd that <strong>the</strong> ON-OFF<br />
times, <strong>the</strong> feedback delays, and a switchscheme sensitivetovariation <strong>in</strong> <strong>ABR</strong> load and<br />
capacity may comb<strong>in</strong>e to create worst case conditions where <strong>the</strong> <strong>ABR</strong> queues diverge.<br />
We have motivated three enhancements to <strong>the</strong> ERICA+ scheme. The modi cations<br />
reduce <strong>the</strong> e ect of <strong>the</strong> variation and allow <strong>the</strong> convergence of <strong>the</strong> <strong>ABR</strong> queues,<br />
without compromis<strong>in</strong>g on <strong>the</strong> e ciency. In <strong>the</strong> next section, we shall exam<strong>in</strong>e <strong>the</strong><br />
e ect of VBR carry<strong>in</strong>g long-range dependent tra c (similar to multiplexed MPEG-2<br />
tra c) on <strong>ABR</strong>, and show that <strong>the</strong> bu er requirements are unchanged.<br />
311