Where to start?

Question 1

You are given a prompt. Start here with these 7 steps.

Answer 1

1. Clarify Requirements / Functionality
2. Scale and Data Estimation
3. System Interface Definition
4. Define Data Model
5. High-level Design
6. Detailed Design
7. Identify and Resolve Bottlenecks

Question 2

1. Clarify Requirements / Functionality

Answer 2

• what is the in-depth functionality?
• what can the users do?
• what data needs to be stored?
• Who/What/When/Where/Why/How?
• which characteristic of Distributed Systems need to be
  prioritized?

Question 3

2. Scale and Data Estimation

Answer 3

• estimate scale of system
• how much storage is needed?
• expected network bandwidth usage?
• this is where you will get information to make informed
  decision on Scaling, Partitioning, Load Balancing, and
  Caching

Question 4

3. System Interface Definition

Answer 4

• define what APIs are expected
• Twitter example:
  • PostTweet(user_id, tweet_data, time_stamp)
  • generateTimeline(user_id, user_location)

Question 5

4. Define Data Model

Answer 5

• clarify how data flows between components
• identify system entities and their interactions
• this is where you will get information to make informed
  decisions on data management: storage, transportation,
  encryption, partitioning

-Twitter example:
- User: userId, Name, Email, LastLogin, etc
- tweet: tweetID, content, numLikes, timestamp, etc
- userFollow: userID1, userID2, etc

• relational or non-relational database?
• what kind of block storage?
• where do you store which type of data?

Question 6

5. High-level Design

Answer 6

• this is where you draw a high-level drawing of the different
  parts of the system and how they interact.
• think about potential bottlenecks, but don’t need to
  resolve them yet.
• Twitter example:
  
  • multiple application servers to serve all read/write
    requests, with load balancers in front of traffic
    distributions.
  • back-end: efficient db that can store all the tweets and
    support large number of reads.
  • distributed file storage system for storing photos and
    data.
• clients –> load balancer –> application servers –> data
  bases and file storage

Question 7

6. Detailed Design

Answer 7

• ask interviewer which major components to dig deeper
  into. typically will be 2-3.
• present different approaches and their pros and cons,
  consider tradeoffs, and explain your choice.
• consider load balancing, caching, SQL v NOSQL, and
  Data partitioning

Question 8

7. Identify and Resolve Bottlenecks

Answer 8

• are there any single points of failures, and how to mitigate
  them?
• do we have enough data replicas?
• do we have enough copies of different services running
  so that failures wont prevent shutdown?
• how are we monitoring the performance of the system?