Yes. Browse Nodes are essentially product categories. So, users search for the same set of categories. There are ~40K browse nodes in each Amazon domain and we have cached most of them already [1]. So, the cache miss is unlikely here.
But, product cache is little complex. We have already cached the top products [2]. But, products are added every day. So, there will be some amount of cache miss there.
Not really! Lets say, user 1 makes an API call to our server to fetch browse node: 1000 (Books). The first time such request is made, our server will make an API call to Amazon using user 1's AWS key. Once we get the response back, we will cache the result and send back the result to user 1. When user 2 makes the request for the same browse node 1000 (Books), we will not hit Amazon server again. Instead, we will get the cached data from our database and send it back to the user 2 without using his/her AWS key. This way the number of API calls made will be less than one per second for a user.
When I was building a price comparison site using Amazon Product Advertising API, I ran into this throttling error frequently: "You are submitting requests too quickly. Please retry your requests at a slower rate."
So, I built a shared cache, so that developers can hit the cache instead of hitting Amazon API every time. That's how I created BrowseNodes. Have you faced the same problem like me? Do you think this will solve the throttling problem for you?