Review cards · 14 cards
Storage
Object, block and file storage, CDNs for big files, and what bytes cost.
Cards
- A file service sends 500 TB a month to users. If data transfer out costs $0.05 per GB, what is the monthly egress bill?
- Why do HLS video segments cache so well on a CDN?
- A paste or file service stores both small facts (owner, expiry, size) and large bodies. Where does each go, and why?
- Large files go to S3 with _____: the file is sent in parts that upload in _____, and a failed part is retried on its own.
- Where do you store user-uploaded photos and videos, and why not on the app servers' disks?
- Object storage offers hot, infrequent-access and archive classes. What do the cheaper ones cost you?
- Clients upload straight to object storage with pre-signed URLs. How should the backend learn that an upload finished?
- Where should a database server keep its data files?
- How can a file sync service avoid storing the same bytes twice?
- Replication protects against lost disks, but not against a buggy job deleting objects, because the delete is replicated too. Protect against that with object _____.
- You store 1 PB of data with 10 + 4 erasure coding (each 10 data pieces get 4 parity pieces). How much raw disk do you need, before headroom?
- How much storage does one hour of video at 5 Mbit/s take?
- How do pre-signed URLs let clients upload large files without the bytes passing through your servers?
- Erasure coding versus 3× replication for durable storage: what do you trade?
More topics
- Estimation 21 cards
- Networking 16 cards
- API design 17 cards
- Caching 21 cards
- Databases 22 cards
- Replication 15 cards
- Sharding 18 cards
- Consistency 19 cards
- Queues 18 cards
- Streaming 18 cards
- Availability 14 cards
- Resilience 16 cards
- Realtime 15 cards
- Data structures 16 cards
- Security 17 cards
- Observability 18 cards
- Coordination 16 cards