Early stress test for Manifold Server
In my previous post I showed how easy it was to stand up a Manifold Server application using an old laptop with 8GB of RAM, and my cable modem with port forwarding to obtain an IP address. If you recall, there was over 250MB of symbolized data and over 100,000 objects.
For the test, I had 10 students hit the site with 10 different click operations (zoom in/zoom out, pan) within 1 minute. As a rough estimate, this means my little laptop was handling 100 requests per minute, or 6000 requests per hour, or 144,000 requests per day. For your average GIS website, I suppose this is a decent amount of traffic.
You can see what my response time looked like when the students were hitting the site:
The results were ok. When I ran the site without any of the students hitting it, my response time was about 1 second per hit. With all the students simultaneously hitting the site, I saw response times between 2 to 7 seconds. I saw the slowest times when I was zoomed out to almost the entire county, and the fastest times when zoomed in really tight. One reason for this was that the rate-limiting-factor was likely due to tunneling the OSM base map through the server. Manifold Server has to wait unti lthe OSM data gets streamed in before it can send it off to the user. I'm going to run another test when simply looking at the data inside of Manifold without OSM.
But if nothing else, this simple shows the ease at which you can build and deploy a map server, while also considering the response time you get with over 100 hits per minute.