It is built on AWS, and we are using PHP and a custom framework for the actual site, and on the backend we're using CutyCapt[1] to generate the thumbnails, which is based on Qt and Webkit.
Yes, we automatically de-dup on the back-end. If the thumbnail you requested exists in the size you want, the information is just returned and you are not charged for the generation. You can tell this happens by looking at the 'created' timestamp in the returned object.
If you request the same thumbnail twice, you will get the same one back twice; however, there is a force_new parameter that can be passed to the generate method that will force it to fetch a new version each time.
We do not have explicit protection right now for DoS attacks, and I thank you for raising that point. In all honesty, the system right now is limited by our ability to handle traffic; you couldn't likely surpass a couple requests a second if you were trying, but it is something we will be thinking about moving forward.
I hope that answers everything for you; if you have any other questions, please let me know.
Yes, we automatically de-dup on the back-end. If the thumbnail you requested exists in the size you want, the information is just returned and you are not charged for the generation. You can tell this happens by looking at the 'created' timestamp in the returned object.
If you request the same thumbnail twice, you will get the same one back twice; however, there is a force_new parameter that can be passed to the generate method that will force it to fetch a new version each time.
We do not have explicit protection right now for DoS attacks, and I thank you for raising that point. In all honesty, the system right now is limited by our ability to handle traffic; you couldn't likely surpass a couple requests a second if you were trying, but it is something we will be thinking about moving forward.
I hope that answers everything for you; if you have any other questions, please let me know.
[1] http://cutycapt.sourceforge.net/