Supporting maximum number of keep-alive connections

2006-03-22 Thread Rajeev Jha
On 3/17/06, Rajeev Jha [EMAIL PROTECTED] wrote: What's your ratio of Apache to Tomcat instances? You may want to look into using squid as a reverse proxy to Tomcat, it is very good at supporting a huge number of concurrent clients without having to spawn a thread or process for each one. I

Re: Supporting maximum number of keep-alive connections

2006-03-22 Thread Remy Maucherat
On 3/22/06, Rajeev Jha [EMAIL PROTECTED] wrote: In our case,the servlet is interfacing to the back-end that sends async events from time to time. As you may have noticed, the HTTP protocol (and the Servlet API) are not designed for this kind of usage. You can try to hack your way through if you

Re: Supporting maximum number of keep-alive connections

2006-03-22 Thread Rajeev Jha
On 3/22/06, Remy Maucherat [EMAIL PROTECTED] wrote: On 3/22/06, Rajeev Jha [EMAIL PROTECTED] wrote: In our case,the servlet is interfacing to the back-end that sends async events from time to time. As you may have noticed, the HTTP protocol (and the Servlet API) are not designed for this

Supporting maximum number of keep-alive connections

2006-03-17 Thread Rajeev Jha
Hi I would like to try out tomcat for new my application. I have used tomcat for quite some time, but the nature of new application is very different from the traditional request-response model. We want to build an application that supports about 1024 keep-alive connections per machine.(2 GB, x86

Re: Supporting maximum number of keep-alive connections

2006-03-17 Thread David Rees
On 3/17/06, Leon Rosenberg [EMAIL PROTECTED] wrote: But pushing dynamic content with squid? I doubt it will work. However, squid is using select reads instead of tomcat's blocking reads, and could reduce the number of threads, but I have seriously doubts with keepalives, have you actually

Re: Supporting maximum number of keep-alive connections

2006-03-17 Thread Leon Rosenberg
On 3/17/06, David Rees [EMAIL PROTECTED] wrote: On 3/17/06, Leon Rosenberg [EMAIL PROTECTED] wrote: Have you tried it? ;) Yeah, and it didn't add any performance. However it helped to keep the thread count low. But it was on 2.4.x kernel, where threads were an issue, on 2.6.x its