摘要 |
Method and system for simulating multiple concurrent clients on a network server to stress test the server. Each of one or more processors has one executable software thread, the send data thread, whereby I/O requests are initiated on behalf of all simulated clients. Each of the one or more processors has another executable software thread, the receive data thread, whereby I/O response data is received from the server on behalf of all simulated clients. A software object, such as a completion port, that is capable of detecting completion of an executable event allows the stress test to function using only two executable threads per processor. The efficiency of the server can be measured as the send data thread initiates I/O requests and as the server responds thereto. The stress test is flexible in that the number of simulated clients and the I/O requests initiated can vary widely. Moreover, many different types of network servers can be tested.
|