What is Interview Forge?

Interview Forge helps you practice mock interviews, follow study plans, track career goals, review learning analytics, and prepare across 36 pages.

Design Google Drive (File Storage & Sync) at Meta

The problem at Meta

Functional: upload/download files, sync across devices, share folders, support offline edits, large file support.
Non-functional: reliable sync, large file support, low-latency listing, eventual conflict resolution.
Scale: ~1B users, ~1TB new files/day, files up to multiple GB.
Consistency: metadata (file tree, versions) is strongly consistent; chunk replication is eventually consistent.

Out of scope (covered in detail below)

Real-time collaborative editing (Google Docs), enterprise admin, OCR/search, and file locking — deferred and discussed in the dedicated section.

At Meta, this problem is framed around move fast, ship impact, scale thinking. Expect Meta-specific scale numbers and a focus on production readiness and the company's engineering bar.

Capacity estimation

Resource	Estimate	Notes
New files/day	~1 TB	Plus revisions
Chunk size	~4 MB	Balance dedup vs metadata
Metadata	~tens of TB	File tree, versions, ACLs
Block storage	~petabytes	Dedup saves a lot here

Architecture overview

Chunked storage + sync

+--------+  upload chunks  +-----------------+  store  +----------------+
| Client |---------------->| Metadata Service|-------->| Block Storage  |
+--------+                  +-----------------+         | (content-addr) |
                                |   ^                +----------------+
                                |   |  chunk refs
                                v   |
                        +---------------+   change events   +--------+
                        | Notification  |------------------>| Client |
                        | Service       |                   +--------+
                        +---------------+

Files become content-addressed chunks; a notification service pushes change events to clients.

Step-by-step request flow

A file is split into fixed-size chunks (e.g., 4 MB); each chunk is content-hashed (SHA-256).
The client asks the metadata service which chunks it already has (dedup); only new chunks are uploaded to block storage.
The metadata service records the file's chunk list and a new version; block storage stores chunks by hash (identical chunks across users share storage).
The notification service pushes a change event to the user's other devices, which fetch only the missing chunks.
On download/sync, the client reconstructs the file from its chunk list, fetching any missing chunks from block storage.

Unlock the full guide

Premium members get every DSA, system design, company and role interview guide — plus AI practice, mock interviews, and analytics.

Unlock with Premium