ExpertTCP - TCP Throughput Testing - GL

ExpertTCP - TCP Throughput Testing(per RFC-6349)818 West Diamond Avenue - Third Floor, Gaithersburg, MD 20878Phone: (301) 670-4784 Fax: (301) 670-9187 Email: info@gl.comWebsite: https://www.gl.com1

Outline Background RFC-2544, Y.1564 (SAM), RFC-6349, SLATCP PrinciplesTCP Throughput Measurement Path MTU Discovery TCP Throughput Inter-Relationships Round Trip Time Measurement Bandwidth * Delay Product Measure TCP Throughput Bottleneck Bandwidth (BB) TCP Congestion Window (TCP CWND) and TCP ScreenshotReceive Window (RWND) Packet Loss Rate Retransmission Schemes (Go Back N, SelectiveRepeat) GL Hardware Platforms2

Performance Testing of Packet / Ethernet Connections and Networks For Predictable Managed NetworksRFC-2544Service Level Agreements ITU Y.1564 (SAM) RFC-6349 (TCP)from Network Providers, a mustUser Experience,Application-Network Sensitive,TCP TuningSAM – Service Activation MethodologyTCP – Transmission Control Protocol3

Packet / Ethernet TestingUser Experience End-to-End Throughput Network ThroughputLatencyPacket LossBack-to-BackJitter4

Typical SLATypicallyPacket Loss0.0005 % to 1%Latency36 to 75 msAvailability99% to 99.9%5

RFC-2544 vs. ITU Y.1564 (ExpertSAM )Both are Connection-less ThroughputLatencyFrame LossBack-to-BackJitter6

RFC-2544 TestingRFC-2544 test application includes the following tests: Throughput - Maximum number of frames per second that can be transmitted without any error. Latency - Measures the time required for a frame to travel from the originating device through the network to thedestination device. Frame Loss - Measures the network’s response in overload conditions. Back-to-Back - It measures the maximum number of frames received at full line rate before a frame is lost.7

ITU Y.1564 (ExpertSAM )Multi-Stream ThroughputLatencyPacket LossJitter8

Testing iceLayer 2/ 3/ 4 SLAIssues like loss, jitterYesYesN/AMulti-serviceLayer 2/ 3/ 4 SLAIssues like loss, jitterNoYesN/ATCP window sizes(CPE issues)NoNoYesExcessive retransmissions due to policingNoNoYes Running RFC-2544, Y.1564 or another L2/L3 layer test is always first step. However, even after these performance tests are passed with good results, end-customers can still complain thatthe “network is slow” and the cause of poor application performance (i.e., FTP, web browsing, etc.). Lack of TCP testing is a turn-up gap because end-customer applications are transported using TCP. Save operating expense costs by eliminating or quickly resolving painful end-customer finger pointing scenarios.9

TCP Principle(Packet Loss and Waiting for ACK Reduces Throughput)10

Major TCP Throughput Inter-Relationships Bandwidth of Applications Latency/Delay of Networks Packet Loss Networks TCP Retransmission Scheme Maximum Transmit Unit of Network Transmit/Receive Windows of TCP # (number) of TCP Simultaneous Connections11

Bandwidth Delay Product (Bits or Bytes)Application and Network are Matched, TCP is TunedB 10 MbpsRTT 50 msB*50 500,000 bitsor 62,500 Bytes65,535 Bytes is maxwindowAchieving maxthroughputBandwidth (B) Bandwidth (bps), Mbps, themaximum rate at which anapplication can transmit orreceive data (the smaller ofthe two). Line rate may beshared among applications.Bandwidth Delay Product(BDP) - measured in bits orbytes (divided by 8), thenumber of bits (or bytes) inthe network that areunacknowledged (intransit), B (bps) * RTT(secs) BDP bits.12

Effect of Increased Network Delayor Smaller Tx or Rx BuffersB 10 MbpsRTT 100 msB*100 1,000,000 bitsor 125,000 BytesBut 65,535 Bytes is maxwindowNOT Achieving maxthroughput, 50% or lessLatency, Delay, Round TripTime (RTT) - in seconds(secs), or milliseconds (ms),round trip time includesacknowledgement delay.TCP Throughput bits/second (bps), millionbits/second (Mbps), One waythroughput (RFC2544,Y.1564), Round-tripthroughput (RFC-6349) is adifferent story sinceretransmissions andacknowledgements areinvolved.13

Effect of Increased Application BandwidthMaximum TransmissionUnit (MTU) - Approx. 1500bytes, max packet size.B 20 MbpsRTT 50 msJitter - Instantaneousvariation in RTT, e.g. if RTT isnominally 100 ms, but variesfrom 80 ms to 120 ms, thenjitter is /- 20ms, or 40ms. Since jitter affects ACKtime, TCP throughput isaffected.B*50 100,000 bitsor 125,000 BytesBut 65,536 Bytes ismax windowNOT Achieving maxthroughput, 50% orlessPacket Loss Rate - Veryimportant factor affecting TCPthroughput, could be as highas 2%.Excess Bandwidth may be used for additional TCP Connections14

Effect of Packet Loss Rate and Retransmission SchemeFor Go Back N retransmission scheme,and if unacked packets is maximum 43or 44, then Packet Loss effects are veryserious!Packet Loss0%0.1 %1%2%TCP Throughput100% 50% 10%0%Probability that one or more MTU packets or ACKpackets is lost is very high!! Can be 1 !!!But for every lost MTU packet or ACK packet, 43retransmissions occur. This results in near zerothroughput.The “slow start phase” results in very few “in flight”packets.15

Effect of Packet Loss Rate and Retransmission Scheme (Contd.)For Selective Repeat retransmissionscheme, and if unacked packets ismaximum 43 or 44, then PacketLoss affects TCP Throughput linearlyfor “low” Packet Loss rates.Packet Loss0%0.1 %1%2%TCP Throughput100% 99 % 95 %? %Probability that one or more MTU packets or ACKpackets lost is very high! But the retransmission onlyaffects the lost packets, not other packets.16

ExpertTCP (RFC-6349 Testing)The TCP Throughput Testing is conducted in 3 steps simultaneously on up to 16 application streams: Path MTU Discovery - What is the maximum packet size that can successfully traverse the network? Round Trip Time (RTT) Measurement - Timestamp based RTT discovery of transmitted packet until acknowledgementpacket arrives from far end. Measure TCP Throughput - Complete measurements per RFC-6349 definitions to provide TCP Throughput results.GL’s ExpertTCP Provides Reports and Graphs of all Results17

GL Hardware / SoftwareExpertTCP 18

Basic SetupTest Configuration of Client and ServerMeasurement Results from Server to Client19

End-to-End Application Performance2Mbpsreplaces all the clients2Mbpsreplaces all the hostsCongestedPacketExpert 1GIPPacketExpert 1GCongestedCongested2Mbps2MbpsMeasure Path MTU RTT TCP Throughput20

ExpertTCP 1G Ports TCP Client and Server will be supported in two different applications. In 1G, Port 2 is used.21

PacketExpert 1G Portable Unit Interfaces 2 x 10/100/1000 Base-T Electrical only 2 x 1000 Base-X Optical OR 10/100/1000 Base-TElectrical Single Mode or Multi Mode Fiber SFP support with LCconnector Optional 4-Port SMA Jack Trigger Board (TTLInput/Output) Protocols: RFC 2544 compliance ITU-T Y.1564 (ExpertSAM) Power: 9 volts, 2.2 Amps Bus Interface: USB 2.0Power(9v, 2A)USB 2.0BackPanelFrontPanel1000 MbpsFiber Interface10/100/1000Mbps CAT-5Interface10/100/1000 MbpsCAT-5 Interface22

PacketExpert mTOP ProbeFront Panel View Rear Panel ViewPortable Quad Port Ethernet/VLAN/MPLS/IP/UDP Tester with 4 Electrical Ethernet Ports (10/100/1000 Mbps) and 2Optical Ports (100/1000 Mbps). Embedded with Single Board Computer (SBC). SBC Specs: Intel Core i3 or optional i7 Equivalent, Windows 10 64-bit Pro Operating System Each GigE port provides independent Ethernet/VLAN/MPLS/IP/UDP testing at wire speed for applications such as BERT,RFC 2544, and Loopback. BERT is implemented for all layers. RFC 2544 is applicable for Layers 2, 2.5, and 3, and Loopback is applicable for Layers 2, 3, and 4.23

PacketExpert High-Density 12/24 GigE Ports mTOP RackPacketExpert SA (PXE112)PacketExpert SA (PXE112) is a 12-Port PacketExpert w/ Embedded Single Board Computer (SBC).SBC Specs: Intel Core i3 or optional i7 Equivalent, 240GBHard drive, 8G Memory (Min), Windows 10 64-bit Pro OS,USB 2.0 or 3.0 Ports, ATX Power Supply. 19" 1URackmount Enclosure (If options, then x 3).PacketExpert SA (PXE124)PacketExpert SA (PXE124) is a 24-Port PacketExpert w/Embedded Single Board Computer (SBC).SBC Specs: Intel Core i3 or optional i7 Equivalent, 240GB Harddrive, 8G Memory (Min), Windows 10 64-bit Pro OS, USB 2.0or 3.0 Ports, ATX Power Supply. 19" stacked 1U RackmountEnclosure (If options, then x 6).24

Step 1. Path MTU Discovery25

Step 1. Path MTU Discovery Path MTU discovery as per RFC 4821 - PLPMTUD - Packetization Layer Path MTU Discovery. DF (Do Not Fragment) bit is set to avoid fragmentation when traversing through network. The algorithm uses TCP retransmit conditions to search for the MTU. Each conclusive probe narrows the MTU search range, either by raising the lower limit on a successful probe orlowering the upper limit on a failed probe. Path MTU is discovered for both directions in case of bi-directional test.26

Step 2. Timestamp based RTT MeasurementTime : 50Time : 100(RTT 150 – 50 100)Time : 150(RTT 170 – 100 70)Time : 170 Timestamp based RTT Measurement (RFC1323). Tx segment includes current time in option field, Receiver echoes timestamp in ACK.27

Step 3. Now Ready to Measure TCP Throughput28

Step 3. Slow Start TCP Throughput Measurement2 x 1500 bytesSlow Start - Initially send two TCP SegmentsIf Acks received, then send double the number ofTCP Segments.4 x 1500 bytesThreshold ReachedContinue doubling until the Receiver “ssthreshold”# is reached, or Acks are not received andTimeout is reached, then halve the send TCPsegments.If Acks are received send TCP segments areincremented by one, until again Timeout isreached, then number of send TCP segments ishalved and the process continues.29

Step 3. TCP Throughput Equilibrium30

Screenshots of Software Operation31

ExpertTCP Main Screen32

Network SetupAll settings configured locally on the client side33

Network Setup (Contd.)Separate Upstream and Downstream bandwidths configurable for asymmetrical path34

TCP SetupSingle TCP connectionMultiple TCP connections35

TCP Setup (Contd.)Upstream/Downstream/Bidirectional Path MTU - run test and discover or user can enter manually. Baseline RTT - run test and find out or user can enter manually. Separate Path MTU/Baseline RTT configuration for Upstream/Downstream directions for asymmetrical paths.36

Status and Results37

Statistics and Periodic ResultsStatistics are updated every second and includes TCP Transmitted Frames/Bytes TCP Retransmitted Frames/Bytes Retransmitted Bytes PercentageThroughput and RTT values are calculated.every second and displayed. Minimum,Maximum and Average Values are displayed.38

Final ResultsIdeal Throughput - the maximum possible TCPthroughput for the given CIR.Ideal Transfer Time - the time taken to transfer the testdata size at the ideal throughput.TCP Transfer Time Ratio - Measure of how muchActual transfer time is greater than the Ideal transfertime.TCP Efficiency - measure of the number of Transmittedbytes compared to the retransmitted bytes.Buffer Delay - measure of how much the RTT increasesduring the actual TCP Throughput test compared to theBaseline RTT.39

Throughput GraphWith 0.1% Packet Loss40

Throughput vs. Retransmitted Frames GraphWith 0.1% Packet Loss41

Multiple TCP connectionsWith 8 TCP connections42

Multiple TCP Connections - ThroughputIndividual Throughput for each connection43

Multiple TCP Connections - ResultImproved Overall Throughput44

Thank you45

Network Throughput Latency Packet Loss Back-to-Back Jitter End-to-End Throughput. 5 Typical SLA . Hard drive, 8G Memory (Min), Windows 10 64-bit Pro OS, USB 2.0 or 3.0 Ports, ATX Power Supply. 19" 1U Rackmount Enclosure (If options, then x 3). . during the actual TCP Throughput test compared to the Baseline RTT. 40 .