IBM G210-1784-00 Manuale Utente

Use the registry editor to set the ListenBackLog parameter located at: 

HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Services\InetInfo\Parameters\ListenBack
Log 

The recommended value is 200. 

The MaxPoolThreads parameter controls the number of threads per CPU in the thread pool available for IIS to run 
Common Gateway Interface (CGI) processes (each process takes one thread). The PoolThreadLimit parameter 
specifies the upper limit for MaxPoolThreads. 

The default thread limit that IIS can create on a machine is twice the number of MB in RAM on a machine (for 
example, a server with 512 MB of RAM is limited to1024 threads). 

Use the registry editor and adjust the following values: 

HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Services\InetInfo\Parameters 
\MaxPoolThreads 

HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Services\InetInfo\Parameters 
\PoolTheadLimit 

The IBM HTTP Server (IHS) is a multi-process, single-threaded server. For more information about tuning IHS, see 
the Hints on Running a high-performance Web server at: 

The configuration takes place in the IHS httpd.conf file located in the confdirectory of the Web server’s installation 
directory. After setting the described parameters, you have to restart IHS. 

Adjust the following IBM HTTP Server properties and parameters to tune your server performance.  

During heavy load, IHS can use a large amount of the server’s processing time to create and destroy httpd child 
processes. For optimum performance, even under heavy load, specify the same value for all four parameters. This 
avoids the time-costly recreation of sub-processes. 

Ignore the appearing error message [error] server reached MaxClients setting. Consider raising the MaxClient when 
restarting IHS. 

A too-low number of child processes is as bad as a too-high number, because more processes also means more 
memory usage and more processor time for managing the processes. You can use the Apache Benchmark (ab) 
program located in the IHS bin directory to test the impact of different parameter settings. 

To simulate 100 concurrent users performing 20 requests to LMS, for example, you can issue the following 
command: 

ab -n 20 -c 100 http://<hostname>/lms-lmm/studentCatalog.do 

In the ab command’s summary, some fields of interest may be the following: 

•  Complete requests 
•  Failed requests 
•  Requests per second