Understanding the key concepts of Load Simulator will help you to get the most out of this tool.
A score is a numerical value. It is measured in milliseconds and represents a server's responsiveness under a given load. Each client module has its own score because each protocol has its own specific user and server interactions. For example, a Microsoft Outlook client and a Microsoft Exchange client might both have a task that sends mail, but they have different scores because they talk to the server in different ways.
Users tend to notice long response times from a server, not more typical response times. Therefore, by default, Load Simulator Log reports 95th percentile response times, not average (arithmetic mean) times. This sets Load Simulator apart from other comparable tools, which generally report average server response times. If you prefer to work with averages or wish to use a different percentile, you can select them using the Load Simulator Log command-line options.
As Load Simulator runs a scenario, it collects the response times in the Lsperf.log file. The Load Simulator Log then sorts and orders the response times and reports the 50th percentile score, the 95th percentile score, the mean score, and the standard deviation.
Load Simulator Log computes percentile scores by picking the time that is higher than 95 percent and 50 percent of the times and reporting them as the 95th percentile and 50th percentile scores. Another way of looking at this is to say that if the 95th percentile response time is X, then 95 times out of 100 the response time is going to be less than X when the server is under the same load.
Some user behaviors are consistent regardless of the messaging client they use. For example, the number of messages a user sends each day can vary from person to person according to their business needs; however, the business needs are the same regardless of client type.
Load Simulator provides three generic (canonical) user profiles with typical sets of loads that you can use for comparison with your own system's loads: Light User, Medium User, and Heavy User.
Note These generic user profiles are provided for comparison purposes only. They are designed only to provide you with a way of comparing data in common scenarios.
These profiles and their corresponding loads are shown in the following table.
| Load | Light User | Medium User | Heavy User |
| Send Mail | 2 | 4 | 6 |
| Process Inbox | 12 | 12 | 12 |
| Browse Mail | 5 | 15 | 20 |
| Calendaring | 1 | 5 | 10 |
| Number of messages in Inbox | 1 | 4 | 9 |
| Number of messages in Deleted Items folder | 1 | 1 | 1 |
| Number of new folders (five messages per new folder) | 20 | 40 | 60 |
A medium user is defined as one who sends 14 messages per day regardless of the type of client. This message load is a combination of new messages, replies to messages, and messages that are being forwarded to other recipients. However, the frequency of tasks performed and the number of messages and folders created during initialization vary according to the profile.
Load Simulator reads in a predetermined set of message files and uses those messages according to a specified distribution. You can tailor the messages by using message files of your own, defining the distributions to use, and saving the message files in the Load Simulator directory.
.sim file Settings specified in Load Simulator are saved in a .sim file. You can create and save files that contain different scenario configurations. When you run Load Simulator it automatically reads the last file used.
.log file Individual action timings collected by Load Simulator are saved in the Lsperf.log file. After you run Load Simulator, use the Load Simulator Log (Lslog.exe) tool to generate scores that are based on the contents of one or more log files.
.out file The Loadsim.out file contains a history of all actions, including errors generated when Load Simulator is running. The messages here are exactly the messages that are displayed when Load Simulator is run. Each error line begins with the word Error for easy searches.
Your experimental topology should designate the servers that are to be used, and the number and types of users involved with each server. The number and size of distribution lists, the Windows NT authentication account, and the public folder hierarchy are also part of the Load Simulator topology.
There are many aspects of your experiment's topology that you should understand and take into account, even though you do not specify them in Load Simulator. They include the configuration of your server hardware and software, the types of networks that connect clients to servers and servers to other servers, the use of connectors between sites, and the use of backbone, hub, or bridgehead servers in the network.
User Groups For each Load Simulator client computer, user groups define how many users are to be connected to each server and what client module is to be used. Each user group is also given a user profile, which can be one of the profiles provided by Load Simulator or a custom profile.
Mailbox and Distribution List Naming Mailboxes are created automatically. During initialization, names are created for emulated users in the Load Simulator scenario. First names consist of four alphabetic characters. Last names consist of a single alphabetic character, followed by the server name, a hyphen, the Load Simulator client module that is to control the user, and the user number for that module on that server. Thus, the name Nbaa Zserver2-exch39 indicates that the mailbox is hosted on server2 using the Microsoft Exchange Server module. This mailbox has the alias of server2-exch39. Mailbox numbers start at zero, so if there are 2000 Microsoft Exchange Server mailboxes on this server, they have the aliases server2-exch0 through server2-exch1999.
Because of this naming scheme and because alias names are limited in length, there is a trade-off between the length of a server name you can use and the number of mailboxes that can be created. For best results, use short server names.
The naming convention for distribution lists consists of three alphabetic characters followed by the letters DL, the site name, a hyphen, and a number (for example, LaaDLmysite-11). The alias for this distribution list is mysite-DL11.
The purpose of the third alphabetic character is to create randomness for both mailboxes and distribution lists. Directory entries are evenly distributed within the alphabet, making directory lookups more realistic.