2

I hope anyone can help me out in this topic, even if it's not a specific programming question. I'm writing a bachelor thesis, where I compare MySQL to MongoDB and I want to write something about Youtube, as the platform has to handle many requests with heavy dataload. The only good resource which I found was this video: Seattle Conference on Scalability: YouTube Scalability

As the conference was in 2007, I can imagine there were some updates regarding to the database.

The last information that I have from this talk is that the thumbnails are stored in a BigTable database and the metadata in MySQL. Are there any changes since then? Where are the videos stored? Is there an entry in the MySQL table, which refers to the stored video?

Thanks in advance for the answer!

kristianp
  • 5,496
  • 37
  • 56
Daniel Blaichinger
  • 348
  • 1
  • 2
  • 13
  • this is a good read on it https://www.8bitmen.com/youtube-database-how-does-it-store-so-many-videos-without-running-out-of-storage-space/ – underdog Dec 16 '19 at 14:33

4 Answers4

6

According to this, youtube still uses mysql: http://code.google.com/p/vitess/wiki/ProjectGoals

kristianp
  • 5,496
  • 37
  • 56
  • they may be using hadoop for performance. site is too large and still fast – kishu27 Apr 13 '12 at 12:19
  • 1
    I'm sure they use a number of different technologies, but it looks like mysql is fairly certainly one of them. Hadoop is not something a google-owned site would use, hadoop is inspired by googles map-reduce. – kristianp Apr 14 '12 at 09:03
  • Thanks for the link! I think I can get some helpful information there. – Daniel Blaichinger Apr 14 '12 at 16:41
3

I am not sure of how things are at youtube but I am in process of developing a similar application for our client. So what we are doing is we are making the use of best of both worlds i.e SQL and NoSQL..

We store the videos on disk and store the path to these videos in MySQL db table. Then we have a separate table which holds the genre and video mapping i.e which video belongs to which particular genre.

Today with vast of pool of user data we are in position to leverage upon these data like we had never been before, so you see things are now way different then 2007 and with the popularity and dependency of people on internet when it comes to sites like you tube we have vast set of unstructured data which if used properly can give you great results. So in our project we store the site admin and reporting stuff like user db, video locations and genre mapping etc in MySQL and store the unstructured data about user interaction in NoSQL database. We then use the NoSQL data to do all the analytics and give appropriate results to the user.

svg
  • 499
  • 5
  • 7
1

They are using mysql with Bigdata. The user information such has who uploaded the file,file information all will be stored in mysql and data will be stored in Bigdata.

Prasad
  • 1,089
  • 13
  • 21
0

I think they are using database that can use FileTable

ruel
  • 1