Scaling a
data-heavy
node.js
application

before-imgafter-img

One of our customers experienced huge scaling problems with a collaborative, heavywrite webapp when being hit by over 1500 concurrent users. Many attempts to improve the situation failed. No matter how many servers were started up the app would become slow even to a point of unresponsiveness. The situation seemed lost.

TheBrain team was called for the rescue!

In 3 months we have researched the problem and:

  • built comprehensive end to end tests around the app (to make sure we are not breaking anything with our performance tweaks)
  • built a custom loadtest solution for Single Page Applications with load scenarios that reused big parts of the code we wrote for the end to end tests
  • decreased the average CPU load with the same number of users by a factor of 10 (!)
  • while cutting the loadtimes of the idle application by half (from even 30 to 15 seconds)
  • All that just in time for the next spike of users. This saved the product, allowed huge savings on infrastructure while making our customers - and their customers - delighted.