Good question. What I read it has to do with scalability. The smaller the thread stack size the more threads you can scale to and the smaller the overall memory footprint that has to be managed. That is what I thought I read someplace.You could say performance and scalability go hand in hand and sometimes one can be misunderstood to mean response time but the question is from a single thread standpoint or from the whole container scalability standpoint? I try to look at both wth the understanding the user experience improves as load increases with the recommended changes. That is how I treat things and I first try in the QA instance the changes and then migrate to production and over a period of a day look for improvements. I have not looked at thread count but then the number of servlets I use is fixed due to the servlet model and the only thread count changes would be what Tomcat provides for incoming requests. Hope that helps and thanks for asking. -Tony
--- On Sun, 7/22/12, Christopher Schultz <ch...@christopherschultz.net> wrote: From: Christopher Schultz <ch...@christopherschultz.net> Subject: Re: Location of Tomcat 7 jvm defualt settings... To: "Tomcat Users List" <users@tomcat.apache.org> Cc: "Tony Anecito" <adanec...@yahoo.com> Date: Sunday, July 22, 2012, 7:17 AM -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Tony, On 7/21/12 3:11 PM, Tony Anecito wrote: > Lots of time I apply best practices that I learn over time or are > recommended by experts (Like Mark Thomas). You do that first as > early in the development life cycle as possible. Funny I guess how > these best practices come from developers, vendors (like Sun and > now Oracle) and DBA's ect. If you can not get your performance to > be the best with light load or expected load (actual users) then > the tests you menton are a waste of time and too late in the > solution life cycle. Some of us do not have million dollar test > systems like HP Performance Center (which I use since I lead a > performance Testing team) or open source tools that support JAX-WS. > So being proactive where you can rather than later is my motto. My > goal was simple and achieved and usage in production was what I > mentioned (1-2msec) and a combination of using leading edge > techniques combined with the old fashioned way of doing things. I'm not sure how you are improving performance by changing the thread stack size. It really sounds like you are grasping at straws at this point. - -chris -----BEGIN PGP SIGNATURE----- Version: GnuPG/MacGPG2 v2.0.17 (Darwin) Comment: GPGTools - http://gpgtools.org Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/ iEYEARECAAYFAlAL/VgACgkQ9CaO5/Lv0PC1lQCfaQn/NK9zz4vhRqRoSR89tT1T w1oAoLYyCaj6LskZ6V4pJnTSzt9h1fsW =PpV7 -----END PGP SIGNATURE----- --------------------------------------------------------------------- To unsubscribe, e-mail: users-unsubscr...@tomcat.apache.org For additional commands, e-mail: users-h...@tomcat.apache.org
