After running this system for a few days, I was able to produce many S3 files as expected, as well as keeping track of each trade in DynamoDB. As I mentioned earlier in this chapter, implementing some view layer wasn't feasible for this example system. However, querying DynamoDB is quite easy, provided primary keys and sort keys are set up correctly. Any view layer that needed to get historical data could easily grab files from S3.
The following screenshot shows listings of S3 files for a given hour. Each file has the format shown earlier with individual prices, along with the pre-computed average for each currency:
Looking one level ...
