Scaling for big projects

[Jahrenski]

Hi!

So I made a test to find out if Restbed can scale well to the size of our projects and I was wondering if this makes sense.

First test is done by publishing a single route and very unscientifically looking at the roundtrip time in chrome devtools to my localhost restbed server.

The average is about 30ms.

Now the scaling test I did was to create 3000 routes in a loop using this:

for (int i = 0; i < 3000; ++i)
{
  auto resource = make_shared< Resource >();

  std::ostringstream ss;
  ss << "/route" << i;

  resource->set_path(ss.str());
  resource->set_method_handler("GET", [](const shared_ptr< Session > session) {
    session->close(OK);
  });

  service.publish(resource);
}

Calling route 999 (999 is the last published route because of the alphanumerical ordering I think?) takes an average of 150ms to complete. Calling route1 averages to 30ms so I guess it uses a set with logarithmic complexity.

This means that there is a delta of about 120ms between first route and route 3000.

So I'm wondering if there are better ways to scale this for a big project.

Like using a unordered_set instead of a set to get the search complexity from log(n) to O(n) in the worst case scenario, and constant in the best case.

I'm also wondering if there would be a way to split routes into groups/modules?

Instead of declaring all routes from the root, you could declare a module such as '/module1/' which would have it's own set of routes with "/module1/" as the root.

Thanks for this project by the way, it seems a good product to use!