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 ...
Figure 8.5: Cell/Packet Drop Points on a TCP/ATM connection result in unnecessary timeouts and retransmissions leading to reduced throughput. Fixed ABR capacity is achieved by not having any VBR source in this case. We simulate the con guration with n = 2, bu er size = 4096 and TBE = 512. In this case, no cells are lost, the CLR is zero and the throughput is 103.32 Mbps. This is the maximum TCP throughput with two sources in this con guration. It can be approximately veri ed as follows: Throughput = 155 Mbps 0.9 for ERICA Target Utilization 48/53 for ATM payload 512/568 for protocol headers (20 TCP +20IP+ 8RFC1577 + 8 AAL5=56bytes) 31/32 for ABR RM cell overhead a fraction (0.9) to account for the TCP startup time ' 103.32 Mbps Figure 8.6 shows graphs of window size, sequence numbers, and ACR for the two sources. Note that the curves for the two sources completely overlap indicating that 269
the performance is fair. Also, the sources use the entire ACR allocated to them. In other words,the TCP sources are rate-limited and not window-limited. Note that given su cient time, the ABR switch algorithm can control the rates of the VCs carrying TCP tra c. We shall quantify this time and corresponding bu er requirements in section 8.14 later in this chapter. Window Size in bytes two-tcp/option-g=1/optionb=295/optiont-g=6/sw_qsize-g=4096/tbe-g=1024/granularity=100/wnd_scale_factor=4/epd_thresh-g=2045/icr=10.0/air=1/xdf=0.5/tdf=0/headroom=1.0 t_threshold=900000.0/maxsrcrate=10.0/ontime=100000/offtime=100000/vbrrate=124.41/t0v=120/a=1.15/b=1.05/qlt=0.8/time_int=1000.0/sw_int=100/dist=1000 / Date:02/03/96 1.2e+06 1e+06 800000 600000 400000 200000 ICR: 10.00 10.00 10.00 10.00 / XRM: 32.00 32.00 32.00 32.00 / Graph: 1 Two TCP : Cwnds Cwnd Size for S1 Cwnd Size for S2 0 0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 Time in milliseconds (a) Congestion Window Rates two-tcp/option-g=1/optionb=295/optiont-g=6/sw_qsize-g=4096/tbe-g=1024/granularity=100/wnd_scale_factor=4/epd_thresh-g=2045/icr=10.0/air=1/xdf=0.5/tdf=0/headroom=1.0 t_threshold=900000.0/maxsrcrate=10.0/ontime=100000/offtime=100000/vbrrate=124.41/t0v=120/a=1.15/b=1.05/qlt=0.8/time_int=1000.0/sw_int=100/dist=1000 / Date:02/03/96 180 160 140 120 100 80 60 40 20 ICR: 10.00 10.00 10.00 10.00 / XRM: 32.00 32.00 32.00 32.00 / Graph: 1 Two TCP : ACRs ACR for S1 ACR for S2 0 0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 Time in milliseconds (b) ACR Figure 8.6: Two TCP Source Con guration, Bu er=4096 cells, TBE=1024 8.10 E ect of Finite Bu ers We now investigate the e ect of smaller bu ers, keeping the ABR capacity xed. The bu er size is set to the product of TBE (512), the number of sources (2), and a safety factor (2), i.e., 2048 = 512 2 2. The remaining con guration is the same as in Section 8.9 i.e., n = 2, TBE = 512 and xed ABR capacity (no VBR source). Since the bu ers are smaller, it is possible that they might over ow before the ABR control loop is set up. We expect some cell loss and reduced throughput due to timeout retransmission. 270
- Page 245 and 246: (a) Transmitted Cell Rate (c) Link
- Page 247 and 248: (a) Transmitted Cell Rate (c) Link
- Page 249 and 250: (a) Transmitted Cell Rate (basic ER
- Page 251 and 252: RM cells. In this chapter, we devot
- Page 253 and 254: low. This tradeo was discovered and
- Page 255 and 256: in the speci cation. b) ACR shall n
- Page 257 and 258: 7.1.6 December 1995 Proposals There
- Page 259 and 260: Figure 7.2: Multiplicative vsAdditi
- Page 261 and 262: is never triggered. However, the PR
- Page 263 and 264: simulation results of bursty source
- Page 265 and 266: 1. The time-based proposal also ind
- Page 267 and 268: The switch maintains a local alloca
- Page 269 and 270: PNI = f0, 1g : f1 ) No rule 5b, 0 )
- Page 271 and 272: after reaching the goal. The time-b
- Page 273 and 274: Figure 7.8: Closed-Loop Bursty Tra
- Page 275 and 276: The algorithm measures the load and
- Page 277 and 278: Medium Bursts Medium bursts are exp
- Page 279 and 280: count-based technique may be insu c
- Page 281 and 282: Figure 7.13: An event trace illustr
- Page 283 and 284: CHAPTER 8 SUPPORTING INTERNET APPLI
- Page 285 and 286: u ering which does not depend upon
- 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: 8.8 Performance Metrics We measure
- 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
Figure 8.5: Cell/Packet Drop Po<strong>in</strong>ts on a TCP/ATM connection<br />
result <strong>in</strong> unnecessary timeouts and retransmissions lead<strong>in</strong>g to reduced throughput.<br />
Fixed <strong>ABR</strong> capacity is achieved by not hav<strong>in</strong>g any VBR source <strong>in</strong> this case.<br />
We simulate <strong>the</strong> con guration with n = 2, bu er size = 4096 and TBE = 512. In<br />
this case, no cells are lost, <strong>the</strong> CLR is zero and <strong>the</strong> throughput is 103.32 Mbps. This<br />
is <strong>the</strong> maximum TCP throughput with two sources <strong>in</strong> this con guration. It can be<br />
approximately veri ed as follows:<br />
Throughput = 155 Mbps<br />
0.9 <strong>for</strong> ERICA Target Utilization<br />
48/53 <strong>for</strong> ATM payload<br />
512/568 <strong>for</strong> protocol headers<br />
(20 TCP +20IP+ 8RFC1577 + 8 AAL5=56bytes)<br />
31/32 <strong>for</strong> <strong>ABR</strong> RM cell overhead<br />
a fraction (0.9) to account <strong>for</strong> <strong>the</strong> TCP startup time<br />
' 103.32 Mbps<br />
Figure 8.6 shows graphs of w<strong>in</strong>dow size, sequence numbers, and ACR <strong>for</strong> <strong>the</strong> two<br />
sources. Note that <strong>the</strong> curves <strong>for</strong> <strong>the</strong> two sources completely overlap <strong>in</strong>dicat<strong>in</strong>g that<br />
269