[Fun] Interesting Fermi Problem
If you’re in consulting or product management, you’ve definitely heard of Fermi problems.
For those unaware, some common questions are like “How many tennis balls can fit into a 747?” or “How many piano tuners are in Chicago?”
The obvious premise of the question is to understand the person’s problem solving and analytical skills.
Note that Wolfram Alpha does provide a nice solution for the tennis problem question in a pretty methodical, precise, and logical way.
Today, I’ll tackle an interesting problem I saw on a random site.
How much storage required to store all of Google Maps?
Seems daunting! We really need to break this problem down to understand really clarify the problem.
Some good questions to ask: Is this for the world? or just US Maps?
After clarifying that, you could talk about your assumptions on what Google Maps specifically stores:
Some common examples: public buildings (DMV), Businesses (Salesforce building), Traffic data (Street-view), Mountains/Terrains (Rocky Mountains), Street Names (Main St), Houses (Urban/Non-Urban), Apartment Complexes.
Now for each one, you could break down the storage.
For example, for Streets, it could be like “Number of Total Cities * Number of Streets per City * Storage for Streets in general”
Some other ones would be tricky but for businesses, it would be like “Number of Total Cities * Number of Businesses per City * Photo Storage space * Text Storage Space”
Eventually, you could add everything up and get something like the below.
(Credit to StellarPeers)
Now, for each of the above, you have to really confirm your assumptions. For example, for the city calculation, you have to talk about NYC as a proxy for a big city, which has about 1000 streets, and it takes 100 KB to store the street name and metadata. If you multiply those you get 100 MB for one big city.
For homes, you can talk about NYC which has 9M people, 3 people per household, so 1M homes. So since each photo is about 4 MB, 100 KB for the home location, you multiply that and you get 12 TB.
Anyway if you add all the assumptions above, you get about 43 PB. Afterward, you have to go back and check if that number is reasonable. Since Google holds about 15 EB total, 43 PB for Google Maps is not absurd.