I like jVisualVM in jdk/bin (1.6+?), also JVM Monitor eclipse plugin works fine for me
we are also doing performance testing of wicket page rendering, seen many comparisons over the web, but cannot get even close to it, on average desktop with 100 threads we get like 2-10 times more memory usage and 10-100 times longer response times, havily dependant on parameters
(compared to pure JSP)
any similar experience?
later spotted, that if all threads are pre-initialized, the response time is just 6-10x longer and memory usage is sometimes even lower (lets assume the same)
after server start i sent as many requests as many threads the server is designed to handle, give some time to rest :) and than it runs smoothly, but i dont like this ...
发布评论
评论（1）
我喜欢 jdk/bin 中的 jVisualVM（1.6+？），而且 JVM Monitor eclipse 插件对我来说工作得很好，
我们也在对 wicket 页面渲染进行性能测试，在网络上看到了很多比较，但平均而言无法接近它具有 100 个线程的桌面，我们的内存使用量增加了 2-10 倍，响应时间延长了 10-100 倍，这很大程度上依赖于参数
（与纯 JSP 相比）
，有类似的经历吗？
后来发现，如果所有线程都预先初始化，则响应时间只会延长 6-10 倍，并且内存使用率有时甚至更低（假设相同）
在服务器启动后，我发送的请求数量与服务器设计的线程数量一样多处理，给一些时间休息:)并且运行顺利，但我不喜欢这样......
I like jVisualVM in jdk/bin (1.6+?), also JVM Monitor eclipse plugin works fine for me
we are also doing performance testing of wicket page rendering, seen many comparisons over the web, but cannot get even close to it, on average desktop with 100 threads we get like 2-10 times more memory usage and 10-100 times longer response times, havily dependant on parameters
(compared to pure JSP)
any similar experience?
later spotted, that if all threads are pre-initialized, the response time is just 6-10x longer and memory usage is sometimes even lower (lets assume the same)
after server start i sent as many requests as many threads the server is designed to handle, give some time to rest :) and than it runs smoothly, but i dont like this ...