Chapter 3: Pastebin

Question 1

Requirements

Answer 1

Functional
1. Users should be able to upload or paste their data and get a unique URL. 
2. Only text data
3. Liks will have a expiry date.
4. Custom alias
Non Functional
1. Highly reliable
2. Highlt Available
3. Realtime with low latency
4. Links should not be guessable.
Extended:
1. Analytics
2. Rest api service.

Question 2

Design Considerations

Answer 2

1. 10 MB paste limit

2. URL size limit

Question 3

Capacity estimation

Answer 3

1. One million pastes per day.
2. Approx 12 pastes per second
3. Approx 60 reads per second

Question 4

Storage estimation

Answer 4

1. 1M * 10kb per day == 10GB per day
2. For 10 years, it will be 36 TB
3. At 70 percent capacity, we need 50TB.
4. Ingress -> 10 KB * 12 == 120KB
5. Egress -> 10 KB * 60 == 600KB
6. Cache for 20% == 10GB

Question 5

API

Answer 5

addPaste(key, data, custom_url, user, paste, expiry_date)

deletePaste(key, url)

Question 6

HLD

Answer 6

I would use a key value store with consistent hashing. He has written two databases, one for metadata and one for pastes, but I don’t think it is good.

Question 7

Application Layer

Answer 7

1. Key generation

2. Duplicates and collision

Question 8

Datastore Layer

Answer 8

1. Key value store
2. Cleanup
3. Partitioning
4. Cache and Load Balancer
5. Security and Permissions