Design a scalable blob storage like Amazon's S3. S3 is a distributed file storage facilitating storage of blob data. In simpler terms, it could be described as the folder on the cloud. The various functions to think about while designing something as robust as S3 are: scaling API requests, scaling storage, durability, handling hot storage nodes, cost efficiency, disk utilization, data redundancy, data corruption, and permission management.
The problem statement is something to start with, be creative and dive into the product details and add constraints and features you think would be important.
While desinging your own S3, account for following features
- seamless scaling of storage nodes
- durability of data even when a node crashes
- handling hot storage nodes
- support for multi-tenancy
- cost efficienct architecture
- maximal disk utilization
- data redundancy across geographies
- handling on-transit data corruption
- file, user, access level permission management
- make your high-level components operate with high availability
- ensure that the data in your system is durable, not matter what happens
- define how your system would behave while scaling-up and scaling-down
- make your system cost-effective and provide a justification for the same
- describe how capacity planning helped you made a good design decision
- think about how other services will interact with your service
- ensure the data in your system is never going in an inconsistent state
- ensure your system is free of deadlocks (if applicable)
- ensure that the throughput of your system is not affected by locking, if it does, state how it would affect
Create a design document of this system/feature stating all critical design decisions, tradeoffs, components, services, and communications. Also specify how your system handles at scale, and what will eventually become a chokepoint.
Do not create unnecessary components, just to make design look complicated. A good design is always simple and elegant. A good way to think about it is if you were to create a spearate process/machine/infra for each component and you will have to code it yourself, would you still do it?
To understand the nuances and internals of this system, build a prototype that
- build a static file server to understand the basics of S3
This is a recommended tech-stack for building this prototype
| Which | Options |
|---|---|
| Language | Golang, Java, C++ |
These are the common pitfalls that you should keep in mind while you are building this prototype
- anything that could fail would fail
- how raw files are served over HTTP
- designing complex systems
If you find this assignment helpful, please
- share this assignment with your friends and peers
- star this repository and help it reach a wider audience
- give me a shoutout on Twitter @arpit_bhayani, or on LinkedIn at @arpitbhayani.
This assignment is part of Arpit's System Design Masterclass - A masterclass that helps you become great at designing scalable, fault-tolerant, and highly available systems.