First Edition May 1992 This is an ASCII text version of the LANPulse manual. This text file contains the basic information needed to operate and understand LANPulse. The manual supplied with a registered copy of LANPulse is bound, contains additional information, graphics, a table of contents and glossary. The following document contains examples of data and reports used in LAN performance evaluation. To illustrate them as completely as possible, the examples may contain names of individuals and companies. All names are fictitious, and any similarity to the names and addresses used by an actual person or corporation is purely coincidental. Requests for copies of this product and for technical information about the system should be made to: Wildon Research PO Box 288 Raleigh, North Carolina 27602 CompuServe ID: 71031,1765 Any comments should also be sent to that address. Wildon may use or distribute any of the information you supply in any way it believes appropriate without incurring any obligations whatever. (c)Copyright Wildon Research 1992 All rights reserved. Overview Capabilities LANPulse was developed to provide LAN administrators with a tool to quantify system performance. You can effectively use LANPulse to perform the following tasks: Performance Monitoring LANPulse provides you with the true bottom line to LAN performance with a graphical representation of throughput, as the user sees it. LANPulse measures true response time from any network node. LANPulse is sensitive enough to detect the effect of a user login as well as demonstrate the performance impact of moving large CAD drawings over the LAN. Configuration Testing LANPulse can be used to compare the effects of equipment and software changes on a LAN. By comparing a before and after graph it is easy to accurately determine impact to performance. Stress Analysis LANPulse, in conjunction with LANLoad, can add a load to your LAN so that its effect can be watched. With this you can simulate the effect of high traffic on any part of the LAN. Modeling LANPulse can be used in a predictive fashion to determine the effect of adding network traffic which does not yet exist. LAN Requirements LANPulse is not specific to any Network Operating System or topology. Any NOS which uses DOS based workstations can use LANPulse. This includes Novell Netware, Banyan VINES, and LANTastic among others. Hardware Requirements IBM compatible computer (386/20 or faster recommended) with at least one 1.5 Mb disk drive space available; one floppy disk (for installation of software); and 640 Kb RAM. LANPulse does not make use of extended or expanded memory. VGA (mono/color) monitor. Note: LANPulse must be able to respond quickly. Therefore do not run LANPulse from a workstation with a clock speed slow enough to become its own performance bottleneck. LANPulse will function but will not provide an accurate upper bound on performance information. A 386/20 machine is generally always fast enough. LANLoad, not having the same speed requirements as LANPulse, may be run from any workstation. Software Requirements LANPulse runs under 2.1 and later releases of DOS. Operation with Microsoft Windows LANPulse is not incompatible with Microsoft Windows but the use of Windows is generally undesirable. Windows does not allow LANPulse to respond quickly enough to accurately report LAN performance. The exception to this is when LANLoad is being used to create a load on the LAN and LANPulse is providing the performance measurement. In this case LANLoad can operate under Windows. The copy of LANPulse measuring the effect of the load should run from DOS. Inventory of LANPulse Files LPPROG.EXE The LANPulse program LANPULSE.BAT Used to start LANPulse. LPLOAD.EXE The LANLoad program (Not included with Demo) LPSETUP.EXE LANPulse setup/configuration program SETDGE.EXE Initializes the LANPulse graphics LPPROC.BAT Batch file executed via Threshold Notification LPINSTAL.EXE Installation program (Not included with Demo) Configuring and Running LANPulse Configuring LANPulse The LANPulse configuration program, LPSETUP, allows you to set which printer and port to use and in what format to send graphical data to the printer. LPSETUP is run automatically following LANPulse installation. To rerun LPSETUP at any subsequent time enter: LPSETUP ÄÙ The Printer Menu From this menu you may select the type printer to use for LANPulse graphs. - HP LaserJet Provides support for the Hewlett Packard family of laser printer. - Postscript Will print to any Postscript capable printer. - Epson Graphics Provides support for dot matrix printer graphics. Any printer which emulates the Epson command set will be supported here. - PCX Image Converts the graphic to a PCX file. You will be prompted for a file name to print to. The resulting PCX file can then be imported into a number of spreadsheet and wordprocessing programs. The Port Menu This menu determines the computer port or disk file that the graphic output will be written to. - LPT1, LPT2 Allows selection between either your first or second parallel port. Note: LPT1 is equivalent to PRN. - COM1, COM2 Allows selection between the first or second serial port. - DISK Prompts for a filename which will be the destination for LANPulse output. The results of this can be printed at a later time. Starting LANPulse Start LANPulse with the command: LANPULSE ÄÙ Command Reference LANPulse Commands The first step in using LANPulse effectively for performance monitoring is to establish an appropriate baseline for your LAN. LANPulse does not diagnose LAN performance. It measures performance and provides you with a view of changes to performance over a selected time span. It is important to understand that a single performance graph tells little. It is essential that you know what is normal for your LAN. Simply put, you must know what is normal for your LAN before you can determine what is not. This section explains the function of each menu command. An understanding of these is necessary prior to using LANPulse for collection of useful data. The Menu System LANPulse menu choices are selected by using the cursor control key to highlight an option and then pressing the Enter key to select it. The right arrow key will open a secondary menu while the left arrow key will close it. The File Menu The File menu offers various choices for retrieving existing data files, creating new files, and exporting to other applications. - Retrieve Prompts for entry of the name of a previously created data file. Once the data is loaded you may Display it on the screen or Export it to another application. If data from a previous test has not yet been saved you will be reminded to save that before you Retrieve new data. Pressing F5 provides a directory listing as shown below. Ctrl-S provides you with several different sorting options. - Save As Prompts for entry of a filename to be used and saves the data. This feature is used to either provide an initial name or to rename a data file. The F5 key will provide a directory listing. When a filename is selected it will appear in the lower left hand corner of the LANPulse screen. - Comment Prompts for entry of several lines of comments in order to provide notes for your future convenience. LANPulse setup parameters are included in Exported data files making that part self documenting. However, you will want to make notes on special conditions present during the test. - Export Creates a copy of LANPulse data in a comma-delimited ASCII format which can be read by Lotus 1-2-3TM, Borland QuattroTM, or Microsoft ExcelTM. This format is very convenient if you intend to manipulate the data in some other way. The filename extension will be .PRN. The Setup Menu The Setup menu contains settings which determine how LANPulse operates. You can determine which nodes will load the LAN, how much load to apply, how often to apply load, when to start, and for how long the test will run. - Transfer From You are prompted to enter a directory name to indicate the point at which LAN loading will originate. Enter a directory name on a local hard disk to create a load from a particular LAN node. A directory on a server disk will create a load leaving the server. Read and write access to this directory is required. - Transfer To You are prompted to enter a directory name to indicate the point at which LAN loading will terminate. You can enter a directory name on a local hard disk to create a load moving to a local node. By using a server directory you create a load ending at the server. You will need write access to any directory selected here. - Transfer Size A secondary menu will open giving you the ability to select the size of the load. Loads ranging from 1Kb to 2Mb are available. Transfer sizes 256Kb and larger, if used in the Continuous mode, will cause a perceptible performance decline on most LANs and are normally used for stress testing only. Transfer sizes 128Kb and less are more often used for performance monitoring. Selecting -Other- will allow you to enter any desired transfer size. You would select this option when you had a specific file size you wanted information on. For example you may want to determine the effect of a workstation to mainframe upload which always references a file of known size. You could simulate the effect by selecting the precise transfer size here. More information on selecting transfer size is available in the sections on Stress Analysis and Performance Monitoring. - Samples (Limited to 100 in the Demo version) You are prompted to enter the number of load transfer to occur. One point on the performance graph will be plotted per transfer. - Frequency A secondary menu opens to allow selection of how frequently the LAN loading is applied. Choices are Continuous, every 5-30 seconds, and every 1-15 minutes. Continuous transfer provides the most accurate performance graph but it also will have the greatest impact on LAN traffic. It is therefore used only for viewing the immediate effects of a LAN event. A 5- 30 second delay is used when a test will run for longer durations (several hours). The 1-15 minute transfer delay is most often used if a copy of LANPulse will be dedicated to constant monitoring of LAN performance. In this fashion LANPulse can gather performance information 24 hours a day without noticeable LAN impact. When a frequency other than Continuous is selected you will be asked to select between Variable and Non-Variable. A variable delay will cause LANPulse to randomly add up to ñ2 seconds to the interval specified. You may wish to use a Variable delay to eliminate the possibility of missing a LAN based event which is occurring at regular intervals. In most cases Non-Variable, which initiates a LAN test at the selected frequency interval, will be fine. - Read/Write A major cause of LAN performance problems relates to reading and writing files on server disks. LANPulse can simulate reading from or writing to a file located anywhere on the network from anywhere on the network (subject to LAN security configuration). A secondary menu opens which allows you to select Read/Write or Read Only operations. Read/Write and Read Only requires you to have write access to the Transfer From and Transfer To directories. - Direction After selecting the Transfer From/Transfer To setting you may want to have the entire test proceed on this direction or have it alternate one for the other. You may select that option here. Uni-Directional adheres to Transfer From/Transfer To settings while Bi-Directional will alternate between them on each test. - Multi-File The Multi-File option allows testing of the effects of disk caches (and their sizes) on LAN performance. For example; if you have a 2Mb disk cache and you are using a 256K transfer size you may want to select Multiple File Names with eight target files. This will cause eight 256K loads to be transmitted thus guaranteeing that the cache will be completely filled. The ability to fill a disk cache has great potential for testing of different cache settings under different loads. Select Single File Name or set Number of Target Files to 0 to turn off the Multi- File option. - Schedule Allows you to specify a time and day to start an unattended LAN test. You will find the feature to be invaluable when testing must happen during your lunch hour or after work. You are prompted to enter a start and end time and date. Use 24 hour time (14:30 = 2:30pm) and be certain that your computer clock is properly set. - Threshold You are prompted to enter a minimum effective transfer rate and duration which will trigger the execution of a customizable batch file. Note: This feature requires that TEXT mode be enabled during a test run since running a batch file in the background is incompatible with graphics mode screen display. For example, you may have decided that when LAN throughput is reduced to 50Kb/sec it will cause user complaints. You will then want to set the minimum effective transfer rate to 50 and then select a duration. The duration is the number of samples that the effective transfer rate has averaged below the effective transfer rate threshold (in this example, 50Kb/sec.) When this situation occurs LANPulse executes the batch file LP_PROC.BAT for you. You can be as creative or a complex as you desire when deciding what to place in LP_PROC.BAT. Perhaps you would have it execute a program which logs the event or perhaps you would send a message to the LAN administrator via your network software. The event is ignored if LP_PROC.BAT does not exist. LANPulse sends three variables to LP_PROC.BAT which you can use however you wish. They are effective transfer rate, current sample number, and warning number. Warning number is a count of messages sent so far. A sample use of these three figures would be as follows. LP_PROC.BAT @echo off netsend "Network Manager" Effective transfer rate of %1 Kb/sec netsend "Network Manager" This is sample number %2 netsend "Network Manager" This is warning number %3 Note: Replace netsend with the appropriate network message command. The Run Menu - Text Provides a real time textual view of a LANPulse test. Use this if your monitor is not capable of providing VGA resolution. The text display shows the current direction of the load transfer by the orientation of the >>> or <<< characters. The Start and Finish time of the current transfer, the Time of the transfer in seconds, the ETR (effective transfer rate), and the sequential Number of the current transfer. This window will continuously update for the duration of the test. - Graph Provides a real time graphical view of a LANPulse test. The PgUp and PgDn keys can be used to rescale the graph on the Y-axis as desired. Up to 230 points can be displayed on a screen. The graph will scroll right to left for each subsequent data point. From this view small changes in LAN performance can be monitored. You can see immediately how your user's applications are affecting LAN performance and can anticipate changes you may need to make. The Display Menu - Text Display, in textual form, the currently loaded LANPulse performance data. This option is disabled if no LANPulse data has been loaded. - Graph Display, in graphical form. The PgUp and PgDn keys can be used to rescale the graph on the Y-axis as desired. The  and keys move the graph left and right along the X-axis. Up to 230 points can be displayed at one time. Alt-S toggles the superimposed mean and standard deviation lines on or off. Alt-C will replace the text at the bottom of the graph with any comments you entered from the File | Comment screen. This allows your comments to appear on the printed graph. Alt-P will print the graph currently appearing on the screen to the printer and port you set in the LPSETUP program. The Exit Menu Exit from LANPulse and return to DOS. Creating a Baseline Graph It is essential you gain a feeling for what is considered normal performance for your LAN. You cannot look at a single graph of LAN performance and tell if it is good or bad without prior information to compare it to. With experience you will be able to look at a graph and tell if your LAN is running as always, or being adversely affected by some new application. You should gather your baseline information during a time of day when your LAN is normally loaded and everything is configured as you wish it to be. Suggested settings for baseline gathering: Transfer From: A server directory Transfer To: Another server directory Transfer Size: Approximate the average size of files transferred via your LAN Samples: 250 (about 10-15 minutes on most LANs) Frequency: Continuous Vary Frequency: Non-Variable Read/Write: Read/Write Direction: Unidirectional Multi-File: Off Schedule: Off Threshold: Off When the test completes you will have a "before" picture of performance which you can use for comparisons later. Either Display the graph and use Alt-P to print or select Export from the File menu, provide LANPulse a filename, and a comma-delimited text file will be created. Then any spreadsheet product can Import this data and allow you to present it in any form you desire. Hint: Always use the same Y axis range on Exported graphs so that differences due to scaling will be eliminated. Remember that 4Mbit token ring looks different from 16Mbit and 10Mbit Ethernet is different from either of those. UNIX based systems have performance characteristics which make them look different from DOS based systems. Keep this in mind and before long you can look at a LANPulse graph and tell the LAN topology, Network Operating System in use, and sometimes even the version of the software being tested. LANPulse is that accurate. Exporting LANPulse Data LANPulse has the ability to print graphs of its own (see Printing LANPulse Graphs) but it also provides a data Export facility. Selecting Export from the File Menu prompts you for the name of a file to export to. LANPulse provides a filename extension of PRN. The exported data is in comma-delimited ASCII format. A sample file, SAMPLE.PRN, with 250 records appears below. "Source = C:\" "Target = C:\WILDON\" "File size = 131072" "Run Date = January 12, 1992" "Samples = 250" "Frequency = Continuous" "Unique Target Names = 0" " " "Test of BitBasher model Z network interface card. " " " " " " " " " " Source to Target" " " "Start","Finish","Secs","KBS","Record" "21:41:01","21:41:01", 0.72, 177.78,1 "21:41:02","21:41:02", 0.55, 232.73,2 "21:41:03","21:41:03", 0.55, 232.73,3 "21:41:03","21:41:04", 0.61, 209.84,4 "21:41:04","21:41:05", 0.61, 209.84,5 "21:41:05","21:41:06", 0.55, 232.73,6 "21:41:06","21:41:06", 0.55, 232.73,7 "21:41:07","21:41:07", 0.55, 232.73,8 "21:41:07","21:41:08", 0.54, 237.04,9 "21:41:08","21:41:09", 0.55, 232.73,10 "21:41:09","21:41:09", 0.66, 193.94,11 "21:41:10","21:41:10", 0.66, 193.94,12 . . . "21:44:42","21:44:43", 0.55, 232.73,245 "21:44:43","21:44:44", 0.55, 232.73,246 "21:44:44","21:44:45", 0.55, 232.73,247 "21:44:45","21:44:46", 0.55, 232.73,248 "21:44:46","21:44:47", 0.55, 232.73,249 "21:44:47","21:44:48", 0.61, 209.84,250 The header information contained here includes: Source: The Transfer From point. Target: The Transfer To point. File Size: The Transfer Size. Run Date: The date of the test. Samples: The number of samples completed. Frequency: The delay interval between each test iteration. Unique Target Names: Number of targets created if Cache Fill was active. Comment: Any comments you entered for the test. Included in the comma-delimited data is: Start: Time of the start of a load test. Finish: Time of the finish of a load test. Secs: Seconds required to complete a test. ETR: Effective Transfer Rate. The LANPulse rating of performance in Effective Kb/sec. This figure is the Y axis on a performance graph. Record: Sequential number of a load test. This figure is the X axis on a performance graph. Once this file is created it can be Imported by a number of different spreadsheet products. Printing a LANPulse Graph The first step towards printing a LANPulse graph is to use LPSETUP to decide on a printer type. The choices include HP LaserJet, Epson, PCX format, and Postscript. See Configuring and Running LANPulse earlier in this document for information on using LPSETUP. Use Display | Graph to place LANPulse data on the screen ready for printing. PgUp and PgDn keys can be used to rescale the graph on the Y-axis as desired. Left cursor and right cursor keys move the graph left and right along the X-axis. Up to 230 points can be displayed at one time. Alt-S toggles the superimposed mean and standard deviation lines on or off. Alt-C will replace the text at the bottom of the graph with any comments you entered from the File | Comment screen. This allows you comments to appear on the printed graph. Alt-P will print the graph currently appearing on the screen to the printer and port you set in the LPSETUP program. The screen background will temporarily change to a black background to indicate that the graph is being sent to the printer. Using LANLoad When testing the performance aspects of a LAN it is often desirable to see how it will perform under stress. LANLoad will generate LAN traffic from the node it is running at while you monitor the LAN performance using LANPulse from another location. As opposed to LANPulse which may not be copied, you may make as many copies of LANLoad as you need. The options in LANLoad are mostly a subset of LANPulse. The main difference being that LANLoad has no reporting capability nor can it save data. It simply as a remote loading mechanism for LANPulse. Performance Monitoring with LANPulse Note: Each of the case studies which follow provide a set of suggested LANPulse settings for you to use. These are just to get you started. You may find a variation of these settings easier for you to use or interpret. As with all complex issues you should tailor the settings to the situation at hand. Have you ever wondered what effect a user's database application has on the LAN?...What happened to performance when you changed versions of your operating system?...A user tells you the LAN is slow today. What does this mean? Can you really compare today with yesterday on a quantitative basis?...What is normal for your LAN? In the following pages let's take a look at some of these questions in the form of case studies. Problem: A user wants to add a database application to the LAN. You are concerned that it may have a performance impact. What is the actual effect on workstation response time when the database application is running? Procedure: If you do not have a baseline graph you should review the section on Creating a Baseline Graph. Make sure the database application to be tested is not running during the baseline test. If you do have a baseline graph you are ready to proceed. Start LANPulse, trying the settings below to get you started, with the database application in operation. Note: Some persons like to start LANPulse before the application, allow LANPulse to run for a minute or two and then start the application to be tested. Allow LANPulse to run for the number of intervals you wish and then save the results. Export the data and print the graph. On the graph you will be able to see exactly when the database application started and ended. Notice the amount by which the graph was shifted downward. The percentage of downward shift shows the level of decreased performance. Suggested Settings: Transfer From: A server directory Transfer To: Another server directory Transfer Size: Approximate the average size of a file transferred through your LAN. Samples: 250 Frequency: Continuous Vary Frequency: Non-Variable Read/Write: Read/Write Direction: Unidirectional Multi-File: Off Schedule: Off Threshold: Off Problem: A user has reported that the LAN seems slow today and you want to know what is wrong. Procedure: Start LANPulse at the normal baseline gathering settings. Allow LANPulse to run for the number of intervals you wish and then save the results. Export the data and print the graph. Compare this graph to your normal baseline graph. Note the differences. On some occasions when a user notices slow performance it is due to a transient load which passes quickly. If you see that the top line of your current graph is consistently lower than the top line of the baseline graph then you have a problem. This may be due to such things as a failing network interface card, or a tape backup being carried out on your portion of the LAN. To track down the problem you can set the Transfer To option to other servers or LAN disks and determine on which the performance is degraded. Suggested Settings: Transfer From: A server directory Transfer To: Another server directory Transfer Size: Approximate the average size of a file transferred through your LAN. Samples: 250 Frequency: Continuous Vary Frequency: Non-Variable Read/Write: Read/Write Direction: Unidirectional Multi-File: Off Schedule: Off Threshold: Off Configuration Testing Have you ever wondered which token ring or Ethernet network interface card has the faster throughput on your network nodes?...Have you ever wanted to see what effect is gained by making you disk cache larger or smaller?...Can you decide which of two disk drives is actually faster in LAN use?...Do you ever have to compare servers side by side? Which brand has the overall better effect on your LAN bottom line -- performance? The following pages outline ways in which you can approach these problems with LANPulse. Problem: Your management has asked you to decide between the faster of several network interface cards. - OR - You must decide which of several computer systems provides overall better throughput for your LAN. Procedure: This type testing is best done on an isolated LAN so that outside interference is eliminated. The perfect situation would be a single workstation attached to a LAN server. If you do not have the luxury of a separate test LAN just be sure to do your testing on a relatively quiet LAN so that user traffic does not affect your results. If you can't get a good time during the day for your test you will want to use the Schedule function to run tests overnight or over the weekend. Add the network interface card or server to be tested to the LAN. Run LANPulse with the suggested settings and graph the results. Change the interface card or try a different server configuration and rerun the test. Look at the graphs to decide which card or configuration provided the greatest benefit. Suggested Settings: Transfer From: Any workstation directory Transfer To: Any server directory Transfer Size: Approximate the average size of a file transferred through your LAN. Samples: 250 Frequency: Continuous Vary Frequency: Non-Variable Read/Write: Read/Write Direction: Bidirectional Multi-File: Off Schedule: Off Threshold: Off Problem: Many network operating systems allow you to adjust the size of a server disk cache. Setting the size of a cache to best fit your usage pattern is an art. It is always a tradeoff between benefitting large files, small files, sequential access databases, or random access databases. LANPulse can help you find an average point that suits your needs. Procedure: This test makes use of the Multi-File feature of LANPulse to saturate any disk cache thereby allowing you to see the effect of changing its size. Before starting, you need to make a decision concerning the average file size that your LAN transfers. Does your LAN mainly handle database traffic? If so, you should use a smaller (1-8Kb) load size. Do you normally use the LAN to load applications from server to workstation? If so, use a midrange (16-256Kb) load size. Do you mainly transfer large CAD drawings from node to node? If so, use one of the larger (256Kb-2Mb) load sizes. Select the Multi-File option. Select Multiple File Names from the next menu. Enter a number for the Number of Target Files which would fill the cache at the current cache size. Example: If your disk cache is 1Mb and your Transfer Size is 256Kb, in order to fill the cache you need to set the Number of Target Files to 4. Set your disk cache to a number of different sizes and repeat the LANPulse test at each size. The classic effect is slow performance with a cache that is too small with performance increasing when the best size is reached and then decreasing with larger cache sizes. Suggested Settings: Transfer From: Any server directory Transfer To: Any other server directory Transfer Size: Approximate the average size of a file transferred through your LAN. Samples: 250 Frequency: Continuous Vary Frequency: Non-Variable Read/Write: Read/Write Direction: Bidirectional Multi-File: On. Schedule: Off Threshold: Off Stress Analysis/Modeling At what point will manufacturer A's network interface card become saturated? How robust is your LAN and how much loading can it take before the users will be affected? The following scenarios make use of both LANPulse and LANLoad to assist you. Problem: You are curious as to how much traffic your new network interface card can stand before it becomes a bottleneck. Procedure: This type testing must be done on an isolated LAN because the point of the test is to saturate the network with traffic. Obviously you don't do that on your production LAN in the middle of a work day. The network interface cards to be tested could be either in the server or the workstation. Note: In this situation you can use LANLoad to provide multiples sources of LAN traffic simultaneously. Run LANPulse with the suggested settings and graph the results. The graphs will show a steep drop in performance when the critical load is reached. Suggested Settings: Transfer From: Any workstation directory Transfer To: Any server directory Transfer Size: 2Mb Samples: 50 Frequency: Continuous Vary Frequency: Non-Variable Read/Write: Read/Write Direction: Bidirectional Multi-File: Off Schedule: Off Threshold: Off Suggested Settings for LANLoad: Transfer From: Any workstation directory Transfer To: Any server directory (multiple copies of LANLoad can all use the same server directory without interfering with its operation) Transfer Size: 2Mb Samples: 50 Frequency: Continuous Vary Frequency: Non-Variable Read/Write: Read/Write Direction: Bidirectional Multi-File: Off Schedule: Off Threshold: Off Advanced Applications You would like to know how well a system will perform under heavy load. Will it degrade gracefully or not? Would you like to determine performance of modems? Have you ever been able to compare separate communication routes or bridges from the basis of throughput? You want LANPulse to monitor your LAN continuously and let you know when things go wrong. The following applications make use of both LANPulse and LANLoad to provide you the answers you need to make an informed decision. Problem: You are curious as to how much traffic your network can handle before performance will degrade. Procedure: This type testing requires the use of LANPulse and LANLoad. LANPulse is to be used as a monitor while copies of LANLoad provide the load generating feature. The copy of LANPulse acting as a monitor could be started with the parameters below. It will run continuously for the duration of this test. LANLoad should be allowed to run for a period of time at each of the Transfer Size settings. By looking at the LANPulse graphical display you will be able to see the changing effect due to the changing load. All loads will be visible as troughs on the graph. You should watch for the point when the top line of the graph falls and fails to regain its previous level. This is the point at which LAN capacity is reached. Suggested Settings: Transfer From: Any server directory Transfer To: Any other server directory Transfer Size: 256Kb Samples: 1000 Frequency: Continuous Vary Frequency: Non-Variable Read/Write: Read/Write Direction: Bidirectional Multi-File: Off Schedule: Off Threshold: Off Problem: You would like to test the throughput of several modems or LAN bridges. - OR - You need to test different communication routes between two points. This test will show you which route has the better throughput which will in turn minimize your connect time and the resulting phone and/or PDN charges. Procedure: Modems, even if rated at the same baud rate, do not have the same effective throughput. Some use data compression to increase performance and some suffer from being overly sensitive to the quality of the data transmission thereby decreasing performance. LANPulse can be easily used to graphically show the sum of many factors affecting the performance of a modem. In this situation you will need read/write access to the server at the remote location and well as at the local end. Suggested LANPulse settings are given below. Run LANPulse using each modem or communication route and graph the results. A simple comparison of the results will be all you need to get a handle on what throughput really is. Suggested Settings: Transfer From: local server directory Transfer To: remote server directory Transfer Size: 16Kb Samples: 150 Frequency: Continuous Vary Frequency: Non-Variable Read/Write: Read/Write Direction: Bidirectional Multi-File: Off Schedule: Off Threshold: Off Problem: You would like LANPulse to monitor your LAN continuously and notify you of performance problems. Procedure: This can be handled by selecting the use of the LANPulse Threshold feature. For example, you may have decided that when LAN throughput is reduced to 50Kb/sec it will cause user complaints. You will then want to set the minimum effective transfer rate to 50 and then select a duration. The duration is the number of samples that the effective transfer rate has averaged below the effective transfer rate threshold (in this example, 50Kb/sec.) When this situation occurs LANPulse executes the batch file LP_PROC.BAT for you. If LP_PROC.BAT does not exist or cannot be found then no action will occur. LANPulse sends three variables to LP_PROC.BAT which you can use however you wish. They are effective transfer rate, current sample number, and warning number. Warning number is a count of messages sent so far. A sample use of these three figures would be as follows. LP_PROC.BAT @echo off netsend "Network Manager" Effective transfer rate of %1 netsend "Network Manager" This is iteration number %2 netsend "Network Manager" This is warning number %3 Note: Replace netsend with the network message command appropriate for your network software. Suggested Settings: Transfer From: Any server directory Transfer To: Any other server directory Transfer Size: 64Kb Samples: 0 Frequency: 5 minutes Vary Frequency: Non-Variable Read/Write: Read/Write Direction: Bidirectional Multi-File: Off Schedule: On (set for time span to monitor) Threshold: On (set ETR and duration)