For our Cloud Service, the top priority for us in our infrastructure is speed and minimizing latency.
To that end we have hundreds of click tracking servers in all Amazon data centers so the redirects are as fast as possible. This infrastructure is globally distributed and extremely scalable.
When there are so many click tracking servers to make sure that the click is always going through the closest one, it becomes impossible to make sure rotations between multiple offers that are rotated are completely even.
For example, 1 click will go to server A and then the next click will go to server B. This can happen very quickly and all these servers don't talk to each other all the time.
We're focused on speed, so making it rotate evenly would take a lot more time and increase the latency.
There's a random factor to the click distribution to ensure that all the offers do get rotated in. Over large numbers, they should be relatively even and matching closer to the rotation weights you set.
Comments
0 comments
Please sign in to leave a comment.