Sena Technologies White Paper: Latency/Throughput Test Of Device .
Latency/Throughput Test ofDevice Servers/Bluetooth-Serial AdaptersSena Technologies White Paper:Latency/Throughput TestofDevice Servers/Bluetooth-Serial AdaptersOctober 30, 2007 Copyright Sena Technologies, Inc 2007All rights strictly reserved. No part of this document may not be reproduced or distributed without writtenconsent of Sena Technologies, Inc.Although every effort has been taken to ensure the accuracy of this document it may be necessary,without notice, to make amendments or correct omissions. Sena Technologies cannot acceptresponsibility for damage, injury, or expenses resulting there from.
Latency/Throughput Test ofDevice Servers/Bluetooth-Serial AdaptersOverviewSena Technologies is a provider of device networking solutions that connect almostany electronic device and equipment to the Internet or Ethernet network using openstandard protocols. Sena Technologies offers highly scalable and affordable end-to-endproducts that provide out-of-box installation functionality to meet industry needs, undera constantly evolving networking environment. In the world of “networked” devices, itis important to understand the basic concepts of the “connection”. This paper willdiscuss the basics of latency and the impact it may have on your network with ourDevices.Figure 1: Device Server ApplicationWhat is Latency?In general, latency can be described as an expression of how much time it takes for apacket of data to get from one designated point to another. Some may also view latencyas a measurement of how much time it takes to send a packet that is returned to thesender or what we call “round-trip time”.In theory latency assumption data should be transmitted instantly between one pointand another, with no delay. Unfortunately, in the real world, contributors to networklatency include: Propagation: The time it takes for a packet to travel between one place andanother. Transmission: The medium itself introduces some delay. Depending on themedium quality, this may create a delay factor.1
Latency/Throughput Test ofDevice Servers/Bluetooth-Serial Adapters Routers and Other Network delays: Each gateway may take time to examineand determine where to send the packets. Other network delays may includenetwork congestion at local switches, network bandwidth limitations, or thenumber of hops in a network required to cross a network. For Example: Themore stop over flights a person makes in a airplane trip from point A to B willincrease the time it takes for that person to reach their destination. Message Size: Depending on the application, the message size can influence thespeed at which the message will be sent. For Example: The heavier the box thelonger it will take for a person to move the box from one room to another. Network Application: Applications can be a source of network delay. Networkapplications include, various operating systems and the actual host application(server configuration, telnet, internet explorer, etc) used in networkcommunication.What is Throughput?Throughput can be described as expression of how much amount of data is sent ortransacted during a defined period of time. It depends on the speed and message size.Sena Technologies Products and Latency / ThroughputAs it was stated earlier, Sena Technologies is committed to providing highlyscalable and cost effective network solutions to the ever changing network andcommunication environment. We here at Sena Technologies view latency as the time ittakes to send one character from a host application to the device server and back againto the host application, or “round trip time” (Due to the variance in different operatingsystem architecture latency measurement times may vary). We also view throughput asthe amount of the data sent to and received from the device server. Testing for latencyand throughput was done in-house by Sena with the full line of Sena products. Figure 2shows how Sena carried out its testing methodology for the device servers and Figure 3for the Bluetooth-serial adapters.Figure 2: Test System Configuration for Device ServerThe test program (written in Visual C ) sends a packet, consisting of one byte ofdata, over an isolated Ethernet network, to a remote device server where the data isechoed back to the host PC. A loop-back plug that connects the RS-232 Tx data pin tothe Rx data pin was used to echo back the data sent by the host pc. Latency time wasmeasured by the time it takes for the one byte of data to be sent by the Host PC to the2
Latency/Throughput Test ofDevice Servers/Bluetooth-Serial AdaptersDevice Server or Terminal Server and back again to the Host PC. For throughput test,the test program sends certain amount of data, measures the time it takes to receive alldata and calculates the received data per second in Mbps. Table 1 displays theparameters that were set to each of the serial ports before testing. From here two typesof latency tests were then conducted, single-port latency tests and full-load latency tests.In the case of throughput, just full-load test was conducted. Full port tests wereconducted by sending data to each and every available serial port at the same time.Sena Technologies also provides Serial/IP software that enables your existing serialcommunication programs that use Windows serial port drivers to be used over thenetwork allowing you to keep using your existing program having to modify yourexisting Windows applications. The second benchmark test which is similar to that ofthe first test provides users a look at how our Serial/IP software plays an impact onSerial to Ethernet latency/throughput. The version used for this test is 4.8.4.Figure 3: Test System Configuration for Bluetooth-Serial AdapterSena Technologies also provides the Bluetooth-serial adapters designed to replaceRS-232 serial cables through the use of Bluetooth connectivity. Latency / Throughputtests of the Bluetooth-serial adapters are conducted in a similar way as those of thedevice servers (Figure 3).Performance TestBaud rateAll the baud rateLatency Test9600bpsData Bits8bitParityNoneStop Bits1bitFlow ControlHardware Flow ControlInter Character Time-out0msCommunication ProtocolTCPOthersData Size : Refer to Table 3: DataSize for Throughput TestNumber Of Iteration : 10000Data Size : 1 byteTable 1: Latency / Throughput Test Parameters3
Latency/Throughput Test ofDevice Servers/Bluetooth-Serial AdaptersPC SPECIFICATIONSProcessorIntel Pentium 4 CPU 3.00GHz (2 CPUs)RAM1024MBOperating SystemMicrosoft Windows XP Professional (5.1, Build 2600)Network AdaptorMarvell Yukon 88E8001/8003/8010 PCI Gigabit Ethernet ControllerNetwork NodesNetwork NodeVendorConnectionNETGEAR FS2108NETGEAR10/100MbpsMarvell Yukon 88E8001/8003/8010 PCI Gigabit Ethernet ControllerMar vell10/100MbpsTable 2: Hardware Used in the TestBaudrateData Size(Kbyte)BaudrateData 966240070Table 3: Data Size for Throughput TestCategoryDevice ServerBluetooth-SerialAdapterProduct NameFirmware VersionSpecificationLS100v1.4.4Ethernet: 10MbpsLS110v1.0.0Ethernet: 10MbpsPS110v1.4.0Ethernet: 10/100MbpsPS410v1.4.0Ethernet: 10/100MbpsPS810v1.2.0.1Ethernet: 10/100MbpsSS100v1.4.4Ethernet: 10/100MbpsSTS400v1.4.4Ethernet: 10/100MbpsSTS800v1.4.4Ethernet: 10/100MbpsSTS1600v1.4.4Ethernet: 10/100MbpsParani SD100v1.1.1Bluetooth Class 1Parani SD200v1.1.1Bluetooth Class 2Parani ESD100v1.1.1Bluetooth Class 1Parani ESD200v1.1.1Bluetooth Class 2Table 4: Products List4
Latency/Throughput Test ofDevice Servers/Bluetooth-Serial AdaptersLatency Test ResultsFinal test results are shown in Table 5, 6 and 7. Table 5 displays the latency time ofa native RS232 serial port. Table 6 displays the latency time of the device servers whenTCP/IP based connection is used. Table 6 also demonstrates the usage of Serial/IP inconjunction with our serial device servers. Table 7 displays the latency time of theBluetooth-serial adapters.ProductNameMethodologyCOM1RS232 Serial PortSingle Port(ms)4-Port(ms)5.51Not Available8-Port(ms)Not Available16-Port(ms)Not AvailableTable 5: Latency Test Result of Native Serial 0STS800STS1600MethodologySingle Not AvailableNot AvailableNot AvailableSerial/IP39.09Not AvailableNot AvailableNot AvailableTCP/IP31.63Not AvailableNot AvailableNot AvailableSerial/IP31.58Not AvailableNot AvailableNot AvailableTCP/IP10.00Not AvailableNot AvailableNot AvailableSerial/IP10.04Not AvailableNot AvailableNot AvailableTCP/IP10.0016.31Not AvailableNot AvailableSerial/IP10.0417.01Not AvailableNot AvailableTCP/IP10.11Not Conducted35.85Not AvailableSerial/IP10.03Not Conducted36.85Not AvailableTCP/IP10.05Not AvailableNot AvailableNot AvailableSerial/IP10.00Not AvailableNot AvailableNot AvailableTCP/IP10.5322.44Not AvailableNot AvailableSerial/IP13.9122.87Not AvailableNot AvailableTCP/IP11.21Not Conducted45.61Not AvailableSerial/IP11.17Not Conducted49.10Not AvailableTCP/IP12.35Not Conducted48.00112.89Serial/IP12.75Not Conducted45.64113.34Table 6: Latency Test Result of Device Servers5
Latency/Throughput Test ofDevice Servers/Bluetooth-Serial AdaptersProductMethodologNameyParani- SD100BluetoothSingle Port(ms)4-Port(ms)8-Port(ms)16-Port(ms)100.96Not AvailableNot AvailableNot AvailableParani- SD200Bluetooth104.77Not AvailableNot AvailableNot AvailableParani -ESD100Bluetooth100.93Not AvailableNot AvailableNot AvailableParani -ESD200Bluetooth103.28Not AvailableNot AvailableNot AvailableTable 7: Latency Test Result of Bluetooth-Serial AdaptersFigure 4: Latency Test Result6
Latency/Throughput Test ofDevice Servers/Bluetooth-Serial AdaptersThroughput Test ResultsFinal test results are categorized by product. Test results display an average baudrate in bps and a percentage of the average baud rate vs. theoretical maximum baud rate.A ratio is also given when compared with that Serial/IP vs. the theoretical maximumbaud rate.The test results of Lite Series are shown in Tables 8 and 9 and Figures 5 and 6.The test results of Pro Series are shown in Tables 10 and 11 and Figures 7 and 8.The test results of Super Series and STS Series are shown in Tables 12 and 13 andFigures 9 and 10.The test results of Bluetooth-Serial Adapters are shown in Table 14 and Figure 11.Lite Baudrate(%)75Not AvailableNot AvailableNot AvailableNot Available150Not AvailableNot AvailableNot AvailableNot Available200Not AvailableNot AvailableNot AvailableNot Available300Not AvailableNot AvailableNot AvailableNot Available600Not AvailableNot AvailableNot AvailableNot Available12001198.7999.901199.0399.921800Not AvailableNot AvailableNot AvailableNot 10.2377388.7367.18230400Not AvailableNot AvailableNot AvailableNot AvailableTable 8: Throughput Test Result of Lite Series(*NOTE: Lite Series does not support such baudrates as 75, 150, 200 and so on)7
Latency/Throughput Test ofDevice Servers/Bluetooth-Serial AdaptersFigure 5: Throughput Test Result of Lite )Serial/IP(%)TCP/IP(%)75Not AvailableNot AvailableNot AvailableNot Available150Not AvailableNot AvailableNot AvailableNot Available200Not AvailableNot AvailableNot AvailableNot Available300Not AvailableNot AvailableNot AvailableNot Available600Not AvailableNot AvailableNot AvailableNot Available120099.9099.8399.9299.921800Not AvailableNot AvailableNot AvailableNot 699.9411520010.2310.0867.1867.17230400Not AvailableNot AvailableNot AvailableNot AvailableTable 9: Throughput Test Result of Lite Series – TCP/IP vs. Serial/IP8
Latency/Throughput Test ofDevice Servers/Bluetooth-Serial AdaptersFigure 6: Throughput Test Result of Lite Series – TCP/IP vs. Serial/IPPro SeriesStatusBaudratePS110AverageBaud ratePS410AverageBaudrate(%)AverageBaud 9410.2230.13Table 10: Throughput Test Result of Pro Series9
Latency/Throughput Test ofDevice Servers/Bluetooth-Serial AdaptersFigure 7: Throughput Test Result of Pro 0.1336.64Table 11: Throughput Test Result of Pro Series – TCP/IP vs. Serial/IP10
Latency/Throughput Test ofDevice Servers/Bluetooth-Serial AdaptersFigure 8: Throughput Test Result of Pro Series – TCP/IP vs. Serial/IPSuper Series and STS 13.96Table 12: Throughput Test Result of Super Series and STS Series11
Latency/Throughput Test ofDevice Servers/Bluetooth-Serial AdaptersFigure 9: Throughput Test Result of Super Series and STS 40029.6327.7827.7727.7727.3233.8313.9616.32Table 13: Throughput Test Result of Super Series and STS Series – TCP/IP vs. Serial/IP12
Latency/Throughput Test ofDevice Servers/Bluetooth-Serial AdaptersFigure 10: Throughput Test Result of Super Series and STS Series – TCP/IP vs. Serial/IPBluetooth-Serial 4628.86Table 14: Throughput Test Result of Bluetooth-Serial Adapters13
Latency/Throughput Test ofDevice Servers/Bluetooth-Serial AdaptersFigure 11: Throughput Test Result of Bluetooth-Serial AdaptersFinal ThoughtsThe benchmark test and test setup, we feel creates a realistic view of performanceunder both real-world configurations under a controlled environment, which give usersand integrators with strict timing designs a good measure of how a Sena device willreact within their network. Although, one must remember that all the latency /performance results reported are an average taken over a number of iterations, realworld results may vary depending on the complexity of various network configurationsout in today’s technologically advanced society.14
Network Application: Applications can be a source of network delay. Network applications include, various operating systems and the actual host application (server configuration, telnet, internet explorer, etc) used in network communication. What is Throughput? Throughput can be described as expression of how much amount of data is sent or
The latency (of endpoint 0) to cache 0 is 100ms. The latency (of endpoint 0) to cache 2 is 200ms. The latency (of endpoint 0) to cache 1 is 300ms. Endpoint 1 has 500ms datacenter latency and is not connected to a cache
data must be restored upon resume thus incurring an expensive transition latency [10]. Using a power mode incurs a sleep and a resume latency. We focus our analysis on resume latency as this, to a large degree, determines the response time of a datacenter using a power state transition policy. Further, if the resume latency is fast enough, long
E7-8890 v3 (18 cores, 2.5GHz, 45MB Cache) 2133 MHz, 512 GB Total: 9 x 2p40GE IntelXL710. 18 x 40GE 720GE !! Latency. 18 x 7.7trillion packets soak test. Average latency: 23 usec. Min Latency: 7 10 usec Max Latency: 3.5 ms. H
4 x Intel Xeon Processor E7-8890 v3 (18cores, 2.5GHz, 45MB Cache) 2133 MHz, 512 GB Total 9 x 2p 40GE Intel XL710 18 x 40GE 720GE !! Latency 18 x 7.7trillion packets soak test Average latency: 23 usec Min Latency: 7 10 usec Max Latency: 3.5 ms Headroom Average vector size 24-27 Max v
Built from fast on-chip SRAM In contrast to off-chip, DRAM “main memory” Average access time of a memory component latency avg latency hit (% miss * latency miss) Hard to get low latency hit and % miss in one structure memory hie
9 Minimum latency is 2**(bInterval-1) USB frames. Default value of 9, will give a latency of 2**(9-1) USB frames. That is 256 frames. As 1 frame is 125us, this gives a latency of 32ms. bInterval can be modified to reduce or increase this latency. Minimum value is 1 and the max is 16. Table 2 - User Configurable Common Configuration - Description
video input and the first pixel of the same video frame going out of the receiver on the video output. The latency is naturally defined in seconds usually in the range of milliseconds for a real-time audio-video system. Video experts also define the latency as the corresponding part of the video stream during that time. The latency is then
Performance Recap Latency is the time it takes to perform a single task Performance(X) 1/ Latency(X) Speedup of X over Y: S Perf(X) / Perf(Y) Latency(Y) / Latency(X) Amdahl's Law limits speedup from optimization Corollary 1: Max speedup bounded by time in optimization Corollary 2: Make the "common case" fast Corollary 3: Parallel speedup limited by time in .
