I'm using ninject with an asp.net MVC5 application and I experiment weird performance behavior only with stacked ajax requests.
Looking to optimize the poor performance of a particular page, I placed a stopwatch between the Application_BeginRequest and Application_EndRequest (let's call it A) and one inside the called method in the controller (let's call it B).
The results are a consistent if I make std http requests (page call) the measurement made by A and B are quite in the same magnitude.
Problem: I display a page with 10 tables, feed by Ajax request. The 4 first requests shows normal behavior (A and B times are respectively 100ms and 80ms) while the 6 other show catastrophic performances: B stays around 80ms while A take up to 10 second.
Is there anythin I can do to avoid such a disturbance, preferably on the server side of the application?
Do not hesitate to ask any information about the setup, I'll be glad to complete the question
It seems that the SessionState was the culprit here, setting it to ReadOnly using [SessionState(SessionStateBehavior.ReadOnly)] drastically improved the performances of the Ajax requests.