Hi I am evaluating JRun 2.3 for our project and wonder if I can take advantage of JRun's multiple JSM (JRun Service Manager) feature for a situation which I will describe. Each production machine on which our servlet application will be deployed is a duo processor machine running Win NT 4.0 server (SP4 or greater) & IIS 4. I am *limited* to two such machines by budget. The servlet application maintains a thread pool. Each thread in the pool hold an address to one and only one dynamically allocated C++ object (legacy stuff) in memory. Communication between each thread and the C++ object is done via JNI. (This might sound like a strange design but there are good reasons which may be too much to discuss now). The next available thread in the pool will process the most current http request. We expect 150 - 200 concurrent http request per web server. Instead of creating one JSM and one instance of the servlet app which manages many threads (50 - 100?) in its pool, I was wondering if I can create multiple JSMs (2 - 4?) each with a servlet app instance that manages less threads (25?) in its pool? I hope that this will scale and perform better. Is this one good way to take advantage of multi JSMs - kinda spread things out among multiple processes/JVMs? I understand that the ideal solution is to get more hardware/machines and have one web server per JSM. However, I am limited to two machines and hope to make the most out of it. If this works out, it will greatly advance the cause of Java (the alternate solution is to write ISAPI filters and C++/COM objects - yucks!!!). Would appreciate any help, opinion or comments. Thanks. -- Josh Lam email@example.com
subject: Taking advantage of JRun's multiple JSMs?