# Design Twitter.
start: 9:55

Twitter is an online social networking service where users post and read short 140-character messages called "tweets." Registered users can post and read tweets

- communication de hieu nen focus on cai gi
- Lead cau chuyen, noi ve nen co nhung chuc nang gi, API nao => input, output => DB
- DB: noi ve table truoc, sau do moi chon loai DB nao => ly do
- ve Diagram, noi ra duoc nhung component chinh => ly do tai sao lai chon => communication giua cac service
- interviewer se pick 1 service va focus on.

**Thieu <- order cua tweets:
Nen dan dat theo user-case:
List up function: <- Tao con thieu gi khong, may co bo sung gi khong.
Scale cua he thong: User?**

Chuc nang chinh:

2 chuc nang chinh:
- Luu tweet vao DB
- Doc tweet tu DB

Cau hoi:
- Co the doc dc tweets cua nhung ai
=> Doc dc twt tu nguoi minh da follow

=> co them chuc nang la
- Follow
- List up nhung nguoi follow
- Dang ki nguoi dung => ko quan tam phan nay

tweeting:
- API tweet: cho phep nguoi dung tao tweet toi da 140 characters va luu vao DB voi userId, tamstamp, contents
- API read tweet:
list up followers => get followers id
=> list ra all tweets dc viet boi followers & myseflt => order by timestamp

- API Follow
luu userId cua minh va following vao table (follow )
- API list up followers
doc tu bang follow ra voi userid la your id

** API: input / output**

Thiet ke DB:

**DB nay thi read nhieu, hay write nhieu? consistency cua no?
**
- users: userId
- tweet: userId, content, timestamp
- follow: userId, followingId
- tweet_for_user <- cached

SQL: Cache: Redis

zset(List[])
key: <- block_memory

key int : +1
List: append(elements)

Tweets?

select followingId from follow where userId = $1 // your id

program => get array following ids

array.push(yourId)

select * from tweet where userId in ({array})

**Chon DB:**
PostgresSQL
- truy xuat nguoc dc voi list follow
- toc do doc nhanh (not sure)
- familiar

1M users follow:
- danh index
- setup master slave => tang server slave

# high level design:

Diagram:
Microservice: function, comnication : sync/async/ (Rest_api , Message_queue)

Main Service:
Insert Tweet into DB

Event: Insert tweet to MQ : user_id, tweet_id, timestamp

1 : timestamp1
2: > timestamp2

MQ: Async
Kafka:

TimelineSevice:

Event(user_id, tweet_id, timestamp)

Get Follower of user: GRPC(user_id) > list of follower

Insert Redis: New Tweet: 1M key insert

Follower service:
List follower of user?

- front end:
- consumer => auto load lai tweet moi
- 140 characters limit
- backend:

- store tweet
- read tweet
- follow user

ca 3 functions vao 1 serve?
LB => n nodes

van de ton tai nguyen do nhu cau cua moi modules khac nhau + he thong la stateless => chon cach chia tung modules ra cac server khac nhau

Q: multi AZ
A: => cloudfront

=> cau hoi order

Q: loat 100 newest tweet

pure query:

select * from tweet where userId in (array) order by timestamp desc limit 100

problem:
- where in => ko indexing?
- scan all records

=> huong giai quyet:
- tweet_for_user
- moi lan tweet: luu lai userId, tweetId, timestamp
=> ko kha thi voi viec write