Fahim ul Haq
When I talk to my friends at Amazon, I can’t help but wish that I’d had the chance to be on their interviewing teams. I’ve already had my own experience conducting System Design Interviews both at Meta and Microsoft, but Amazon just plays a different ballgame. Namely, leadership principles play an especially significant role in their candidate evaluations (an emphasis that I personally respect).
If you have a System Design Interview coming up at Amazon, it’s important to prepare well and know what to expect. So today I’ll discuss System Design concepts, strategies, and common System Design problems to prepare for — as well as how to display leadership principles in Amazon’s System Design Interview.
Let’s get started.
The basics of the Amazon System Design Interview
The core purpose of this interview is to assess your approach to designing a system, critical thinking and creativity during design, the reasoning behind choosing components, and your ability to evolve the system based on requirements.
What you’ll be asked
System Design questions are rather vague and open-ended. They can prompt you to design a range of simple to complex, large-scale systems.
Some commonly asked System Design questions at Amazon include:
- How to design a parking lot system
- How to design a web crawler
- How to design a payment system for Amazon’s Kindle
- How to design a rate limiter
- How to design a video streaming service like Amazon Prime
- How to design a system for an e-commerce website like Amazon
Each design requires different design choices, depending on the constraints and requirements of the problem. While you get a broad question, you’re expected to ask clarifying questions from your interviewers to fully understand these nuances of your given problem.
What is expected from you
Amazon’s interviewers don’t expect you to have experience working with large-scale systems. But they do expect you to know System Design fundamentals, and to leverage them to design a scalable system.
Because there are various possible solutions to a given problem, it’s more important for interviewers to see a sound thought process as you complete a design — rather than a single solution itself. System Design Interviews aren’t a heads-down session; you need to communicate and clearly explain the reasoning behind your decisions.
On top of all this, you’ll need to present a solution to the design problem within a specific time, no more than 45 minutes. To hit all the right points in the right timeframe, having a confident strategy (and even a handy framework) is essential.
And what’s unique to Amazon is that displaying leadership principles are a priority for them, even in the System Design Interview. You have the opportunity to demonstrate these principles with how you engage in your interview. For example:
- Graciously taking feedback displays Earn Trust
- Discussing how to improve your design displays Insist on the Highest Standards Offering to discuss a component deeper displays Dive Deep Let’s now discuss how to prepare for the System Design Interview at Amazon.
Preparing for the Amazon System Design Interview
What to study
So what should you learn to prepare?
To start, you should learn the basic components or building blocks of System Design. Each building block covers the details related to a specific functionality that can later be part of a complex system. You can combine them to create any large-scale system you want — like Lego pieces. You can start by exploring basic System Design concepts that you should be able to discuss during the interview. (This guide to System Design Interviews covers most of the basic concepts.)
Next, it’s very important to understand the functional and non-functional requirements of different systems you may be asked to design.
You should pay attention to trade-offs. Trade-offs are compromises that need to be made in your system’s abilities. There is no such thing as a perfect system that satisfies all requirements, and certain trade-offs have to be made in every design.
Finally, you have to apply all this knowledge by practicing designing real-world problems. On that note, let’s discuss how to approach a solution to any design problem.
How to design large-scale systems
The most optimal approach to a problem is to break it down into smaller subproblems and solve them step by step. The System Design problem can best be answered by breaking it into the following steps:
- Clarify the goals.
- Determine the scope.
- Define a high-level design.
- Start simple, then iterate.
- Consider a relevant DSA.
- Describe the trade-offs.
It is a good start, but solving problems by following only these steps would be challenging would be challenging. For example, the interviewer can ask about detailed design, estimating capacity, defining a schema for data storage, evaluating requirements, etc.
We need a more practical or systematic approach that helps to answer all the questions an interviewer might ask. Let’s discuss it.
A practical approach to System Design
There is no universal formula for approaching a System Design problem, but a practical approach can be handy in such situations and help you remember key steps to follow while approaching a solution.
We defined such an approach and named it RESHADED, as illustrated below:
Remember: Our comprehensive System Design course details the basic components and implementation of the RESHADED approach to example design problems.
Let’s discuss Amazon’s most frequently asked System Design problems. We can start with simple System Design problems and gradually move toward complex problems.
Common System Design problems by Amazon
How to design a parking lot system
You’ll design a system allowing users to reserve parking spaces at a parking lot. The users can pay to confirm their reservation and cancel the parking if needed. Let’s start with the requirements.
Functional requirements:
- List parking spaces: The users should be able to list parking spaces for their vehicle type.
- Reserve a space: The users can reserve a space of their desire.
- Payment: The service should allow users to pay a fee to confirm their slot.
- Cancel reservation: The users should be able to cancel their reservation.
Nonfunctional requirements
- Consistency: The system should be consistent so that no two users are allowed to reserve a slot simultaneously.
- Availability: The system should be available for a seamless user experience.
The high-level design
The system should accept users’ requests to list the available parking slots. If a user selects a slot to reserve, the system should reserve a space for a specified time, allowing users to make the payment to confirm the slot. Once confirmed, the data should be written to the databases. We can also use replicas for strong read consistency.
Trade-offs
A higher latency is possible for strong consistency, as the system needs to ensure the reservation after the payment is processed. This is because we introduce a coordinator between the nodes, such as a payment and reservation service, that employs a protocol such as a two-phase commit, introducing a delay but assuring consistency.
How to design a payment gateway
You’ll be required to design a system like Stripe that processes users’ payments through credit or debit cards for their merchants, such as an e-commerce store or the parking lot system. The system would be able to process the payments and generate receipts.
Functional requirements
- Payment: The system should allow users to initiate payment processing.
- User data management: The system should allow users to add or update payment information.
- Invoices: The users should be able to generate invoices.
- Transaction details: The system should provide details once the transaction succeeds.
Nonfunctional requirements
- Consistency and data integrity: The system should provide strong consistency and integrity of transactions.
- Availability: The system should be highly available for a seamless user experience. It includes a failover mechanism, redundancy, and backup for quick recovery.
- Scalability: The system should scale to the ever-increasing number of users and payment methods.
- Security: Security must be prioritized to secure users’ sensitive information.
The high-level design
Clients interact through a merchant’s interface with a payment service. The payment system forwards the user and payment information to the payment gateway, which interacts with the risk evaluator service. After clearance, the payment gateway requests the issuer bank to perform the transaction from the user’s account to the merchant’s. Payment service responds with a receipt of payment to the user after a successful transaction.
Iteration to design
The system’s flow is understandable, but this high-level design is not as scalable to the increasing number of users. We should introduce the following into our system:
- We should use microservices for different operations such as customer management, invoices, transaction histories, payout processing, balance, etc.
- We must use a pub-sub service to facilitate communication between different services.
- We should define backups and data replication strategies.
- The load balancer should play its part in balancing the load between all services.
Trade-offs
The use of microservices, replication for backup, pub-sub, etc., ensure availability and scalability but incur extra cost and overhead to manage each service separately. Moreover, as the system communicates with the risk evaluator and issuer’s bank through a payment gateway, it can affect the latency, and we can bear that at the cost of consistency, data integrity, and availability.
Our chapter discusses the payment system in detail to help you better understand the workflow and trade-offs.
How to design an Amazon Prime Video
You can be asked to design a streaming service such as Amazon Prime, YouTube, or Netflix. Since the approach to designing all three services is the same, we’ll go with Amazon Prime as of now.
The streaming service should store and stream thousands of videos seamlessly to millions of users.
Functional requirements
- Search videos
- Upload videos in different formats
- Stream videos of different qualities
- Process payments
Nonfunctional requirements
- Availability: The system should be highly available for a seamless user experience.
- Scalability: The system should scale to support a seamless experience for millions of concurrent users.
- Security: Security is essential as we’re using a payment service, and only users with paid subscriptions are allowed to use the service. Moreover, we need to allow only authorized users to upload content.
- Latency: There should be no/minimum buffering time to stream a video.
The high-level design
The video files should be uploaded to the service, which passes them through encoders and transcoders to convert to different formats. We store formatted videos and metadata in storage for streaming.
Iteration to design
Now, we should reiterate our high-level design to meet requirements. We use a blob store to store large video files and an SQL server to store users’ and videos’ metadata. Moreover, to achieve low latency, it is crucial to place videos in the nearest locations of the users. For that, we distribute videos to the content delivery networks (CDNs) to achieve low latency while streaming.
Do you think this is the complete design, or can we expand it? What about search and payment functionality integration? How would you handle the time to upload larger video files? Can you think of techniques to support availability? Are CDNs good enough for low latency? Think of a solution and then explore the System Design of YouTube to gauge your understanding.The other key System Design problems asked by Amazon are explored in the Top 14 System Design Interview questions for software engineers as listed below:
- Design a chat service
- Design a ride-sharing service
- Design a URL-shortening service
- Design a social media newsfeed
- Design a social message board
- Design Instagram
- Design a file-sharing service
- Design Google Docs
- Design an API Rate Limiter
- Design a web crawler
- Design a proximity service
- Design typeahead
- Design Google Maps
- Design a video streaming service
Tips to ace the System Design Interview
To succeed in the Amazon System Design Interview, here are some common practices to follow:
- Communicate: Always communicate your thought process. It is possible that the interviewer might perceive your decision as wrong, you should explain and justify the why behind that.
- Don’t make assumptions: Avoid getting into unnecessary details or making assumptions without valid reasons.
- Practice is key to success: The key to success is clearly understanding basic concepts and practicing different real-time design problems.
- Start small and reiterate: Always start small while approaching a solution, reiterate, and improve to scale the system according to requirements. Don’t be shy and ask clarifying questions.
- Display leadership principles: From Think Big to Customer Obsession, demonstrating these principles is essential at Amazon.
Remember, there are no right or wrong answers in System Design Interviews, and every answer is right if you can justify it. When you get into detailed design, with your critical thinking ability, you will understand if you need to change an earlier design. Your approach will highlight your ability to adapt and modify the design to accommodate requirements and can increase your chances of landing a job at Amazon.
Next steps
Acing the System Design Interview requires practicing and thorough study. This guide is only a beginner to mastering the art of System Design Interviews. You need an in-depth understanding of more concepts and design problems to prepare for a real interview. You can get hands-on practice with our AI mock interviews to gain an in-depth understanding of real-world design problems.
Check out the following resources to read and understand the basic concepts of distributed systems:
Distributed Systems for Practitioners
Moreover, to understand the concepts of building blocks or System Design components and to crack example design problems, the following course can be of great help:
Grokking Modern System Design Interview for Engineers & Managers
To ace the API design phase during a System Design, the following course is filled with in-depth knowledge about designing APIs:
