Ajax responses slower to controller in physically different file

Question

When trying to speed up some ajax calls on one of our MVC pages, I came across some strange behavior that I cannot explain. I have several ajax calls made every N seconds to poll some statistics.

It seems that the ajax calls made to the controller in a physically different file are significantly slower than the similar calls made to the controller in the same physical file as in the case where the view comes from.

See my simplified examples:

Situation 1: Only 1 file

Foocontroller.cs

namespace FooBar.Areas.FooArea.Controllers { [SessionState(System.Web.SessionState.SessionStateBehavior.ReadOnly)] public class FooTestController: _BaseController { public JsonResult GetStats() { try { var req = new RestRequest() { Method = Method.GET }; req.AddHeader("Content-Type", "application/json"); req.AddHeader("Accept", "application/json"); req.AddParameter("apikey", /*APIKEY*/); var client = new RestClient(/*STATSURL*/); var response = client.Execute(req); if (response.StatusCode == HttpStatusCode.OK) return Json(new { success = true, content = response.Content }); else return Json(new { success = false }); } catch { return Json(new { success = false }); } } public JsonResult GetAgents() { var req = new RestRequest() { Method = Method.GET }; req.AddHeader("Content-Type", "application/json"); req.AddHeader("Accept", "application/json"); req.AddParameter("apikey", /*APIKEY*/); try { var client = new RestClient(/*AGENTSURL*/); var response = client.Execute(req); if (response.StatusCode == HttpStatusCode.OK) return Json(new { success = true, content = response.Content }); else return Json(new { success = false }); } catch { return Json(new { success = false }); } } } public class FooController : _BaseController { // VIEW OF THE PAGE MAKING THE AJAX REQUESTS public ActionResult Index() { Title = "Home"; return View(); } } } 

Situation 2: 2 separate files in one folder

Foocontroller.cs

 namespace FooBar.Areas.FooArea.Controllers { public class FooController: _BaseController { // VIEW OF THE PAGE MAKING THE AJAX REQUESTS public ActionResult Index() { Title = "Home"; return View(); } } } 

FooAjaxController.cs

 namespace FooBar.Areas.FooArea.Controllers { [SessionState(System.Web.SessionState.SessionStateBehavior.ReadOnly)] public class FooAjaxController: _BaseController { public JsonResult GetStats() { try { var req = new RestRequest() { Method = Method.GET }; req.AddHeader("Content-Type", "application/json"); req.AddHeader("Accept", "application/json"); req.AddParameter("apikey", /*APIKEY*/); var client = new RestClient(/*STATSURL*/); var response = client.Execute(req); if (response.StatusCode == HttpStatusCode.OK) return Json(new { success = true, content = response.Content }); else return Json(new { success = false }); } catch { return Json(new { success = false }); } } public JsonResult GetAgents() { var req = new RestRequest() { Method = Method.GET }; req.AddHeader("Content-Type", "application/json"); req.AddHeader("Accept", "application/json"); req.AddParameter("apikey", /*APIKEY*/); try { var client = new RestClient(/*AGENTSURL*/); var response = client.Execute(req); if (response.StatusCode == HttpStatusCode.OK) return Json(new { success = true, content = response.Content }); else return Json(new { success = false }); } catch { return Json(new { success = false }); } } } } 

In both situations, ajax calls are made from jQuery as follows:

JQuery

 $.ajax({ url: // URL TO ACTION DEPENDING ON SITUATION, type: "POST", dataType: "json", cache: false, success: function (result) { if (result.success) { var content = JSON.parse(result.content); console.log(content); } } }); 

Now the response time from ajax requests for both situations is as follows: on the left, situation 1 is displayed, and situation 2 on the right:

So, as you can see, the average time of a call made in GetStats() and GetAgents() in situation 1 is 52.8 ms and 53.8 ms, respectively.

However, in situation 2, the average call time is 486.8 ms and 529.9 ms .

Now my question is: how can it happen that ajax calls made to actions are on average almost 10 times slower when these actions are in the controller in a physically different file than when these actions are in the controller that has the same physical file as The file that displays the view first?

Is it because the file containing the action to render the view is already loaded and stored in memory, while a separate file, as in situation 2, opens and closes each time the action is called? Or is something more sinister going on?

Answer 1

You shared most of the code, but not _BaseController.cs

• For each request, MVC receives a controller using dependency injection.
  [spec] MVC can store the last used controller, returning it once several times. However, if you switch controllers, perhaps every time he creates a new one.

• Perhaps the default _BaseController constructor has very slow code - perhaps database queries. This sounds unlikely, but happened in my experience.

Taken together, this will lead to a slowdown that you describe.