How much storage is required to store all of the images on Google Maps?
+6 votes
in Estimation by (725 points) | 561 views

1 Answer

+5 votes

I can take a stab at this.

Are we talking about user submitted images of key spots to Google Maps which includes both flat and spherical images, icons? Will this also include all the images shot by 360′ street view, and user submitted photos (both flat and spherical)? I’ll assume, yes.

Calculating type of assets and file sizes –

1. Assets used –
a. Flat Photos – A 12 Megapixel photo is anywhere from 5-20 MB based on compression. We can go with 5MB assuming there are some compression algorithms.
b. Spherical Photos ( 360 ) – Google Maps now has street view 360 maps on many roads in N. America. But these are limited to roadways, key landmarks & national parks. Possibly around 15MB each
c. Icons, Text – average around 5KB

Calculating Google Maps coverage –

Google Maps has maps covering most of the world both urban & rural. So I would start by calculating the habitable area in the world.

We know that radius of earth is 6500 km’s. Surface area of earth would then be 4 x Pi x r (squared) ~ 500 Million sq. km’s. Earth is about 25% land, so land area is 125 Million sq. km’s. Assuming about 40% of this is habitable, we end up with 50 Million sq. km’s.

Now let’s assume Google has covered at least 70% of this habitable area = 35 M sq. km’s.

Calculating volume of assets –

1. Per 10 sq. km we can assume there are an average of at least 50 points of interest (gas stations, restaurants, parks, homes, etc.) which demand image assets.
2. Total number of national parks and landmarks in the world, let’s assume about 100,000 that have some sort of spherical media uploaded.
3. Assuming MAU’s on Google Maps is around 1 Billions users. Let’s assume about 10% upload at least 20 images to Maps. This gives us 2 Billion images uploaded
4. Per 10 sq. km we can assume there are an average of about 10 roads and 20 homes which demand icons or copy assets
a. Assuming 70% of roads have spherical media

With math above we arrive at these numbers on approximate number of assets stored in the cloud –
1. Points of Interest Images = 50 * 35 M sq. km/10 sq. km = 175 M images
2. 100,000 Landmarks with 100 spherical images each = 10 M spherical images
3. 2 Billion images uploaded by users
4. Roads & Homes
a. 70% * 35/10 * 10 = 25M spherical images
b. 35/10 * 20 = 7M images
5. Icons & Copy – assuming this as a total of all the above (points of interest + uploaded images + landmarks roads) = 2.875 B

Storage –
Flat images – 2.875 B * 5 MB = 14,375 TB
Spherical images – 35 M * 15 MB = 525 TB
Icons & Copy – 2.875B * 0.005 MB = 14.3 TB

Total storage = ~15K TB or 15 PB

Hi Chandra,

Thank you for posting your answer. Great step by step breakdown of the answer to smaller pieces. I think the approach is right and you did the right job of breaking the total number of pictures to three main buckets.

My only feedback is around how you’ve calcualted the number of pictures in each bucket. Although the interviewer is not looking to test your skills in guessing the right numbers for things such as number of points of interest per certain size of land or number of parks in the planet, they’d like to see how you can find a way to make an educated guess that’s backed by some reasoning. You want to break down the equation to the point where you can finally reach a number that’s reasonably guessable / justifiable. For that reason, I would have tried to break down the equations further to estimate the number of pictures per certain size of land. One way could be to estimate number of pictures for a small town (e.g. town of 2,000 people) and then mulitply that number by an x factor to represent the whole planet with 7B population. To estimate number of pictures in a town of 2,000 people, you can estimate the length of the roads (based on estimated size and number of houses) and estimate number of points of interest based on potential number of parks / malls / etc per 2,000 people. Check out my answer for number of street view images in another exercise.

You can also go further in your estimations and break down number of images per land size based on type (e.g. buildings, houses, points of interests, shops, etc) and then estimate the number for each type.

Overall, I think the structure and answer was right. I would have just done more calculation to estimate the number of photo’s in a small size of land / population before expanding to a larger number. Hope it helps
Thanks Bijan for the feedback!
Your name to display (optional):
Privacy: Your email address will only be used for sending these notifications.
To avoid this verification in future, please log in or register.
Your name to display (optional):
Privacy: Your email address will only be used for sending these notifications.
To avoid this verification in future, please log in or register.

Related questions

+5 votes
3 answers
0 votes
2 answers
asked Mar 20 in Estimation by emily (21 points) | 760 views
0 votes
3 answers