Upload
others
View
0
Download
0
Embed Size (px)
Citation preview
CROSS-LAYERCROSS-LAYERLATENCY-AWARE AND -PREDICTABLELATENCY-AWARE AND -PREDICTABLE
DATA COMMUNICATIONDATA COMMUNICATIONAndreas Schmidt
Apr 07th, 2020
Anywhere on Earth
Doctoral Colloquium
復
1
CYBER-PHYSICAL SYSTEMSCYBER-PHYSICAL SYSTEMS
2 . 1
DISTRIBUTED CYBER-PHYSICAL SYSTEMSDISTRIBUTED CYBER-PHYSICAL SYSTEMS
2 . 2
A REAL-WORLDA REAL-WORLDDISTRIBUTED CPSDISTRIBUTED CPS
Proprietary SolutionEnhanced Shock Burst (ESB)
Our Open SolutionWi-Fi, IP, PRRT
Publication-in-Progress [BSGPH20]0:00 / 0:41
2 . 3
2008 2008
Cyber Physical Systems:Design Challenges [Lee08]
"Passing of time is inexorable"
Most abstractions insufficiently capture timing
Edward A. LeeUC Berkeley
2 . 4
LATENCY-*LATENCY-* AWARENESS AWARENESS
Know what latency you requireKnow what latency you causeShare this with others
PREDICTABILITY PREDICTABILITY
Built upon awarenessImprove confidenceShare confidence with others
2 . 5
STATE OF AFFAIRS IN NETWORKINGSTATE OF AFFAIRS IN NETWORKINGThe Internet
most dominant interconnecting communication systemsbest effort, i.e. no predictable timing (by design)
Link: Transmission latencies known, but not exposed
IP: No timing at all
Transport Layer:TCP: UDP: RTP:
Let's go back to proprietary, closed networks? No! Why?
2 . 6
Internet Architecture [Cla18] provides
INTEROPERABILITYINTEROPERABILITY
2 . 7
RESEARCH QUESTIONRESEARCH QUESTION**
CAN WE MAINTAINCAN WE MAINTAIN INTEROPERABILITY INTEROPERABILITY AND AND AT THE SAME TIME AT THE SAME TIME
ACHIEVEACHIEVE LATENCY-AWARENESS AND -PREDICTABILITY? LATENCY-AWARENESS AND -PREDICTABILITY?
* as you might assume, this is a question too big for a PhD thesis
2 . 8
TODAY'S RESEARCH QUESTIONTODAY'S RESEARCH QUESTION**
CAN WE BUILD CAN WE BUILD TRANSPORT LAYERSTRANSPORT LAYERSWITHWITH
LATENCY-AWARENESS AND -PREDICTABILITY?LATENCY-AWARENESS AND -PREDICTABILITY?
*
2 . 9
OUTLINEOUTLINE
PRRTPRRT X-LAPX-LAP X-PACEX-PACE TTSTTS OUTLOOKOUTLOOK
2 . 10
PRRTPRRT[SH16a, GGG+19]
3 . 1
WHAT IS THE PROBLEM?WHAT IS THE PROBLEM?Full reliability infinite time [Sha48, Fei54]
Most applications...
... have (some) time constraint
... can tolerate (some) missing data
(Today's) transport protocols do not consider that
⇒
3 . 2
FLAVOURS OF RELIABILITYFLAVOURS OF RELIABILITY
Full Partial
3 . 3
PRRT PROTOCOLPRRT PROTOCOLPredictably Reliable Real-time Transport
Fundamental work in [Gor12]
Redesign for control applications since 2015
3 . 4
PRRT'S PLACE IN THE ISO MODELPRRT'S PLACE IN THE ISO MODELDefault
Control/Multimedia App
PRRT
UDP
IP
Any MAC
In GeneralControl/Multimedia App
PRRT
Any Layer withProcess-to-ProcessDatagram Delivery
3 . 5
PRRT CHANNEL MEASUREMENTPRRT CHANNEL MEASUREMENTParameter Approach
Round-trip Time NTP Algorithm [MMBK10]
Delivery Rate IETF Dra� [CCYJ17]
Loss Rate & Correlation Statistics over success-failure sequences
Selective ACKs with feedback
Measurements exposed via API
3 . 6
PRRT'S RECEIVE MODESPRRT'S RECEIVE MODES
3 . 7
PRRT'S FUNCTIONSPRRT'S FUNCTIONSError Control
Adaptive Hybrid Error Coding [Gor12] (not covered in this talk)
Congestion ControlBased on BBR ideas [CCG+16]Utilization & measurementLatency-avoidance
Rate and Flow ControlLater: X-Pace
3 . 8
EVALUATIONEVALUATION
[GGG+19]
3 . 9
USABILITYUSABILITY Policy
Publicly available at
MIT licensed
Technology
C implementation & API
Python API (via Cython)
Rust API (via bindgen)
prrt.larn.systems
3 . 10
http://prrt.larn.systems/
X-LAPX-LAP[RSH+17, RSH+18]
4 . 1
WHAT IS THE PROBLEM?WHAT IS THE PROBLEM? CPS demand predictable latency and timing
Cross-layer, intra-host profiling is requiredvalgrind, wireshark, packetdrill... insufficient
Microsecond regime is increasingly relevant [BMPR17]
4 . 2
STAMPSSTAMPS Wall-Time
CLOCK_MONOTONIC
expensive to capture (≈ 70ns)
Processor Cyclesrdtsc
cheap to capture (≈ 10ns)non-trivial relation to wall-time
4 . 3
WHAT IS X-LAP?WHAT IS X-LAP?X = Cross, Lap = time to do one round
X-Lap is a
cross-layer, intra-host, timing & latency
analysis tool
download at xlap.larn.systems
4 . 4
http://xlap.larn.systems/
STAMPS IN X-LAPSTAMPS IN X-LAP
e.g.: ChannelTransmit
[ms]
[1]
↑t = 5 ms ↓
c = 7 156 294
4 . 5
X-LAP IN ACTIONX-LAP IN ACTION
4 . 6
TRACES DATA FORMATTRACES DATA FORMATSN ... ChannelTransmit_T ChannelTransmit_C ...
42 ... 5.012ms ...
43 ... 5.029ms ...
... ... ... .... ...
7 156 294
47 961 303
4 . 7
RUNTIME TRACING CODERUNTIME TRACING CODE
XlapStamp_Cycle(LinkTransmitStart, pkt->seqno);
XlapStamp_TimeValue(ChannelTransmit, timestamp, pkt->seqno); XlapStamp_CycleValue(ChannelTransmit, cyclestamp, pkt->seqno);
XlapStamp_Cycle(LinkTransmitEnd, pkt->seqno);
void send_packet(PrrtPacket* pkt) {1 pace(); // delays sending of data using cross-layer pacing2 wait_for_free_congestion_window_space();3 compute_next_send_time(); // for cross-layer pacing4 char[] bytes = serialize_packet_to_bytes(pkt);5
6 struct timespec timestamp;7 uint64_t cyclestamp;8 send(sock_fd, bytes, & timestamp, &cyclestamp);9
1011
track_outstanding_packet(pkt);1213
}14
4 . 8
CYCLE TO TIMECYCLE TO TIME
4 . 9
PACKET TRACEPACKET TRACE
4 . 10
TRACE JITTERTRACE JITTER
4 . 11
MULTISERIES CORRELATIONMULTISERIES CORRELATION
4 . 12
X-PACEX-PACE[SRGP+19]
5 . 1
WHAT IS THE PROBLEM?WHAT IS THE PROBLEM? Bufferbloat [GN12]
Unpaced Paced
5 . 2
PACEPACE PACINGPACINGintentionally delaying
a transmissionP , [P ] = time
unit of work
5 . 3
CROSS-LAYER PACINGCROSS-LAYER PACINGevery step considers
and adapts its pace to thebottleneck pace
PACED SYSTEMSPACED SYSTEMS
i ∈ [0 : n − 1]P (i)
P (btl)
S is paced iff
∀i, j ∈ [0 : n − 1] :
i < j ⇒ ≥P (i) P (j)
5 . 4
PROPAGATE PACESPROPAGATE PACES
5 . 5
REDUCE 無駄REDUCE 無駄*** muda or waste, originally described by Toyota Production System [Ohn88]
Inventory: only a delivered message is useful
Waiting: a message loses value over time
Overprocessing: as-fast-as-possible causes losses
5 . 6
X-PACE IN PRRTX-PACE IN PRRT
5 . 7
CROSS-LAYER APICROSS-LAYER API
send_sync()Passive
socket.btl_paceActive
5 . 8
EVALUATION APPROACHEVALUATION APPROACHPRRT with X-Pace
vs.
Optimized TCP Variants (CUBIC, BBR) NODELAY, QUICKACKlow_latencySNDBUF & RCVBUF lowwrite() not send()
5 . 9
INTERNET SCENARIOINTERNET SCENARIOtraceroute to 79.199.28.123 (79.199.28.123), 30 hops max, 60 byte packets 1 vlan-herfet-neu.nt.uni-saarland.de (134.96.86.1) 0.400 ms 0.358 ms 0.420 ms 2 c65eb36-win.net.uni-saarland.de (134.96.6.54) 0.328 ms 0.314 ms 0.384 ms 3 cr-dui1-te0-5-0-2-4.x-win.dfn.de (188.1.241.185) 8.687 ms 8.688 ms 8.680 ms 4 cr-fra2-be16.x-win.dfn.de (188.1.144.178) 9.618 ms 9.732 ms 9.727 ms 5 ffm-b12-link.telia.net (213.248.97.40) 9.226 ms 9.563 ms 9.191 ms 6 ffm-bb3-link.telia.net (62.115.142.46) 10.069 ms 9.761 ms 9.704 ms 7 ffm-b4-link.telia.net (62.115.120.6) 9.802 ms 9.792 ms 10.024 ms 8 dtag-ic-319284-ffm-b4.c.telia.net (213.248.93.187) 10.374 ms 10.175 ms 10.17 9 91.23.246.213 (91.23.246.213) 13.230 ms 13.252 ms 13.234 ms 10 * * * ... 30 * * *
5 . 10
RESULTSRESULTS
5 . 11
TRANSPARENT TRANSMISSIONTRANSPARENT TRANSMISSIONSEGMENTATIONSEGMENTATION
[SH16b, SH17b]
6 . 1
WHAT IS A PROBLEM?WHAT IS A PROBLEM? Small receiver buffer flow-limited communication ⇒
6 . 2
TRANSPARENT TRANSMISSIONTRANSPARENT TRANSMISSIONSEGMENTATIONSEGMENTATION
Relay: Transport-Layer Performance-Enhancing-Proxy [ ] RFC3135
6 . 3
https://tools.ietf.org/html/rfc3135
TTS IN ACTIONTTS IN ACTION
6 . 4
IDEAL RECEIVER BUFFERIDEAL RECEIVER BUFFERderived from a theoretical flow control model
= ⋅BrelayRT T1
RT T2Brecv
Λ = = + 1Reff,TTS
Reff,E2E
RT T1
RT T2
6 . 5
EVALUATIONEVALUATION= 4 KiB, RT = 20 ms, RT = 5 msBrecv T1 T2
⇒ Λ = 5, = 4 ⋅ 4KiBBrelay
6 . 6
BENEFITS OF TTSBENEFITS OF TTSLossy Last MileHigh Reordering due to JitterInsufficient Receiver Buffer
6 . 7
OUTLOOKOUTLOOKe.LARN, [SGPH20]
7 . 1
FUTURE WORKFUTURE WORK
Energy-Awareness
StatisticalShaping
MulticastSupport
7 . 2
WRAP-UPWRAP-UP
NETWORKED CPSNETWORKED CPS
PRRTPRRT[SH16a, GGG+19]
X-LAPX-LAP[RSH+17, RSH+18]
X-PACEX-PACE[SRGP+19]
TTSTTS[SH16b, SH17b]
OUTLOOKOUTLOOKe.LARN, [SGPH20]
8 . 1
QUESTIONSQUESTIONS
8 . 2
APPENDIXAPPENDIX
9 . 1
PUBLICATIONS PUBLICATIONS[SH16a] Schmidt, Andreas; Herfet, Thorsten: “Approaches for Resilience- and Latency-Aware Networking“, InternationalSymposium on Networked Cyber-Physical Systems (NetCPS, Poster Session), Munich, Germany, September 2016
[SH16b] Schmidt, Andreas; Lange, Tobias; Herfet, Thorsten: “Low-Latency Multimedia Streaming using OpenNetworking Environments“, International Conference on Computer and Communications (ICCC), Chengdu, China,October 2016
[SH17b] Schmidt, Andreas; Herfet, Thorsten: “Transparent Transmission Segmentation in So�ware-Defined Networks”,IEEE Conference on Network So�warization (NetSo�), Bologna, Italy, July 2017
[RSH+17] Reif, Stefan; Schmidt, Andreas; Hönig, Timo; Herfet, Thorsten; Schröder-Preikschat, Wolfgang: “X-Lap: ASystems Approach for Cross-Layer Profiling and Latency Analysis for Cyber-Physical Networks”, 15th InternationalWorkshop on Real-Time Networks (ECRTS RTN), Dubrovnic, Croatia, June 2017
[RSH+18] Reif, Stefan; Schmidt, Andreas; Hönig, Timo; Herfet, Thorsten; Schröder-Preikschat, Wolfgang: “∆elta:Differential Energy-Efficiency, Latency, and Timing Analysis for Real-Time Networks“, 16th International Workshop onReal-Time Networks (ECRTS RTN), Barcelona, Spain, July 2018
[GGG+19] Gallenmüller, Sebastian; Glebke, René; Günther, Stephan; Hauser, Eric; Leclaire, Maurice; Reif, Stefan; Rüth,Jan; Schmidt, Andreas; Carle, Georg; Herfet, Thorsten; Schröder-Preikschat, Wolfgang; Wehrle, Klaus: “Enabling wirelessnetwork support for gain scheduled control“. 2nd International Workshop on Edge Systems, Analytics and Networking(EdgeSys), Dresden, Germany, March 2019
[SRGP+19] Schmidt, Andreas; Reif, Stefan; Gil Pereira, Pablo; Hönig, Timo; Herfet, Thorsten; Schröder-Preikschat,Wolfgang: “Cross-layer Pacing for Predictably Low Latency”. 6th International IEEE Workshop on Ultra-Low Latency inWireless Networks (ULLWN), Paris, France, April 2019
[SGPH20] Schmidt, Andreas; Gil Pereira, Pablo; Herfet, Thorsten: "Predictably Reliable Real-time Transport Services forWireless Cyber-Physical Systems", IFAC World Congress, Berlin, Germany, July 2020
9 . 2
REFERENCES REFERENCES[Sha48] Claude Elwood Shannon. A Mathematical Theory of Communication. 1948.
[Fei54] Amiel Feinstein. A new basic theorem of information theory. 1954.
[Lee08] Edward A. Lee. Cyber Physical Systems: Design Challenges. In Proceedings of the 11th IEEE InternationalSymposium on Object and Component-Oriented Real-Time Distributed Computing (ISORC), 2008.
[MMBK10] D. Mills, J. Martin, J. Burbank, and W. Kasch. Network time protocol version 4: Protocol and algorithmsspecification. RFC 5905, RFC Editor, June 2010. http://www.rfc-editor.org/rfc/rfc5905.txt.
[GN12] Jim Gettys and Kathleen Nichols. Bufferbloat: Dark Buffers in the Internet. Communications of the ACM, 55(1):57– 65, 2012.
[Gor12] Manuel Gorius. Adaptive Delay-constrained Internet Media Transport. PhD thesis, Saarland University, 2012.
[CCG+16] Neal Cardwell, Yuchung Cheng, C Stephen Gunn, Soheil Hassas Yeganeh, and Van Jacobson. BBR: Congestion-based congestion control. ACM Queue, 14(5):50, 2016.
[BMPR17] Luiz Barroso, Mike Marty, David Patterson, and Parthasarathy Ranganathan. Attack of the Killer MicrosecondsCommunications of the ACM, 60(4):48–54, March 2017.
[CCYJ17] Yuchung Cheng, Neal Cardwell, Soheil Hassas Yeganeh, and Van Jacobson. Delivery rate estimation - IETFDRAFT, 2017.
[Cla18] David D Clark. Designing an Internet. MIT Press, 2018.
9 . 3
IMAGE SOURCES IMAGE SOURCESEdward Lee
9 . 4
https://ptolemy.berkeley.edu/~eal/