On Jan 30, 2010, at 03:16 , Leonardo Ribeiro Queiros wrote: [...] > The access of the web service had a high performance when only ONE > client was using it. However, We had problems when we had more than > only one client at the same time using the web service. For ONE > client the time for request was 20 ms, but for FIVE clients it was > 10 s and for 50 clients it was 40 s.
Sounds like you have a nasty contention in your resource implementation. What's behind it? > We are really worried about the performance for more than one > client because we have been working on this web service for more > than ONE YEAR. We need to solve this problem urgently to try to put > the web service in production. > We are using the Restlet 1 and Simple conector. The clients > aplications request the Web Service with AJAX to autocomplete a > TextArea with a list of terms. The response of Web service's XML is > small - 4K. Why are you using the Simple connector? I'd strongly suggest switching to Jetty. Where is the data coming from for the XML? Is the nasty contention there or in how you're using whatever it is? > After a lot of research, We suspect that the bad performance > happens because the Restlet 1 is synchronous and if We try to > migrate to the Restlet 2 We maybe solve the problem. To do this > migration my staff will spend a lot of time so, I would like to know > if someone recomends this procedure to solve the problem with the > bad performance. If someone say yes, we will do this doesn't matter > how much time it involves, but if someone say no, we will not wast > more time. Well, I'm not going to encourage or discourage a switch to the latest Restlet implementation since I haven't kept up with the latest releases. I have v1.0.x code that was in production at Krugle and took some serious hammering in load and worked just fine -- I deployed used Jetty and appropriate settings. Take care, John ------------------------------------------------------ http://restlet.tigris.org/ds/viewMessage.do?dsForumId=4447&dsMessageId=2443745
