-
User is able to share videos
-
User is able to watch videos on the FYP
-
User is able to watch vdeos on "Following" Page
-
User is able to Comment on the videos
-
User is able to like the videos
-
User is able to authicate
-
User is able to create account
-
User is able to modify profile and setting configuration. Private account. Change name, change location.
-
The username cannot change
Extra
- User is able to livestream
- Chat in the livestream
- send stickers in livestream
1B Users
-
Scalability
-
Availablity. 99.999% (Redundancy, Replication ) Computer: Mutliple server, node, cluster Storage: Eventual consistency
-
Robustness
-
Reliability
other:
- Cost
- Deployability
- Sustaibility, modularityh, extendibility
CAP Theorem:
Sacrifice Consistency for Availability
RDBMS. ACID properties.
Instagram, Netflix, Facebook
NoSQL -> Cassandra,
Banq Consistency, Stock Market
LMAX System (One node) No Partition - Stock market
Capapcity Estiamtes:
- 1B Users
User Data: -Username -DOB -CreationDate -LastLogin
1B x 100KB = 100 TB Mistake
-Statistics -Comments -Likes
Video Data:
1 video per user per day on average?
1B vidoes per day shared
10 MB?
10 PB per day shared
7 Gb per minute 100 MB Per second
Video MetaData:
- Title
- Description
- Creation data
- CreatorId
- VideoId Mistake
Followers:
UserId Follower UserId
DataLake with Statistics Video watch time etc. Likes watch etc. ML - Analytics
Grocking - System Design Alex XU - System design interview Youtube Videos
excalidraw.com/ alidraw.com/
https://excalidraw.com/#room=272723ef74217500ed11,b4e2uDda6aLSoa4cm9aThA