Cloud storage looks uniform from the outside: files sit “in the cloud”. Technically, very different systems hide behind that phrase, and the differences determine which tools work, which guarantees apply, and what changing providers costs.
Matching commands
Ready-to-run Storage commands for PowerShell and the Unix shell, with examples to copy.
Articles on Storage (5)
- Sep 4, 2026 log4j2 disk space When the log fills the disk: properly limiting log4j2 RollingFile, using totemomail as an example
- Jul 26, 2026 Paperless with Cloud Service Running Paperless-ngx with Limited Storage: Offload Documents to a Cloud Service
- Jul 26, 2026 Proton Drive & Linux Proton Drive on Linux: Status as of July 2026
- Jul 26, 2026 Rclone in Docker Running Rclone mounts reliably in Docker
- Jul 26, 2026 Testing Rclone Testing Cold Storage with Rclone: A Practical Test Plan
Object storage versus file system
Object storage (the S3 model) has no real folders, only objects with keys, metadata, and checksums in flat buckets. It scales almost without limit, is inexpensive, and forms the basis of most backup and archive services. File-based services (drive models) behave like a remote file system with folders, versions, and shares, and are designed for end-user clients. The distinction explains many quirks: renaming in object storage is a copy operation, change detection relies on checksums or timestamps, and not every service even exposes a documented API for third-party tools.
Sync, copy, mount: three access patterns
Copy transfers data in one direction and leaves existing data in place; sync makes a destination match a source exactly and in doing so also deletes whatever is missing at the source; a mount presents remote storage as a local file system. The patterns differ fundamentally in risk: a sync command with source and destination swapped deletes data, and a mount simulates local semantics that the remote service cannot always deliver (latency, locking, partial writes). Tools such as rclone implement all three patterns uniformly across many services and add checksum verification (rclone check) and client-side encryption (the crypt backend).
Encryption: where the key lives
Three levels need to be distinguished: transport encryption (TLS, standard everywhere), server-side encryption (the provider encrypts but holds the keys), and client-side or end-to-end encryption (only the customer holds the keys, and the provider sees random data). Zero-knowledge services and client-side tools trade convenience for this: server-side full-text search, previews, and web-based editing cannot work on encrypted content by design.
Storage classes and retrieval
Providers tier prices by access pattern: hot for active data, cool for infrequent access, cold/archive for retention with retrieval fees and sometimes hours of lead time. For archive data, cold storage is cheaper by orders of magnitude; the price is that restoring costs money and time and therefore needs to be tested before anyone relies on it.
