You’ve heard of data lakes. How about a data ocean?
In the Ocean's Trilogy, Danny Ocean and friends attempt to cheat casinos. In the 2007 Ocean's Thirteen, the heroes are faced with a new challenge: an intelligent machine called the Greco Player Tracker, a security system for the new Bank Hotel and Casino. In this three part blog series, we'll learn how to build a Greco Player Tracker (GPT) ourselves.
The GPT “has a hell of a brain.” The machine can:
analyze biometric input from every player
analyze every permutation of every wager in every seat
determine whether someone is cheating
The movie claims the it works “in a field of exabytes” and does so in real time. If you've taken Introduction to Big Data, you know an exabyte is a lot of data! So can we build a GPT using the MapR Converged Data Platform?
Number of Players
Let's assume the Bank Hotel and Casino is the same size as the Bellagio Resort in Las Vegas. The gaming space in the Bellagio is 116,000 square feet. Assuming we need 20 square feet per person, we have a maximum capacity of around 5,800 people. The casino typically won't be operating at maximum capacity, but we want our GPT to be able to handle this.
The Bellagio Casino has 2,300 slot and video poker machines and other table games. Assuming each of these play a hand, roll, or round once per minute, and each generates around 100 kilobytes of data and metadata, we're looking at 350 GB of data daily at full capacity.
Heart rate, body temperature, sweat, and pupil dilation can be used to determine whether a player is excited from a legitimate win, or nervous from cheating. Heart rate, temperature, and sweat can be monitored with a wristband. Theme parks like Disney World already issue wristbands, although they do not track biometric data.
Our players could keep track of their winnings, hotel key, and bar tab with such a wristband. The addition of sensors like the ones in some fitness trackers allow us to track heart rates and body temperatures. These generate mostly integers and associated metadata. Even with 5,800 players, this will only produce a few gigabytes daily.
The Bellagio has around 2,000 security cameras. Recording 30 FPS at three megapixel resolution, these cameras will generate 170 megabytes per second each. This means our GPT will need to stream 340 GBPS, and store over seven petabytes of new data daily. Once we have this video data, we can use facial recognition to identify players and measure their pupil dilation. This, combined with other biometric data, will tell us if they're cheating.
In our next post, we'll use these parameters to build the Greco Player Tracker with the MapR Converged Data Platform.