Comparison · 2026-07-08 · 8 min read
The phrase “cloud storage” now covers a genuinely wide range of designs, and the differences matter a lot depending on what you are storing and why. Here is a practical, non-tribal comparison of the main approaches, and where a network like Walrus sits among them. If you are earlier in the process and just want the map of what exists, start with Cloud Storage Alternatives.
Centralized object storage (S3, GCS, Azure Blob)
The default for most of the industry. Excellent tooling, very high durability within the provider, and deep ecosystem integration. The trade-offs are structural: one company controls access and pricing, egress fees can be steep, and you are trusting a single operator’s continuity and policies. For most day-to-day workloads it is the pragmatic choice — but it concentrates trust and control. We dug into the cost side of that in Egress Fees and Vendor Lock-In.
IPFS: addressing, not permanence
IPFS is a content-addressing and distribution protocol — files are referenced by a hash of their content rather than a location. That is powerful for integrity and deduplication, but IPFS by itself does not guarantee your data stays hosted; something (you, or a paid “pinning” service) has to keep it available. It solves addressing beautifully and leaves durability as a separate problem to fund.
Arweave: pay once, store forever
Arweave’s model is a one-time payment for permanent storage, funded by an endowment designed to pay for storage indefinitely. It is a strong fit for data that should truly never disappear — public archives, provenance records. The permanence is the feature and the constraint: you commit up front, and the model is optimized for “forever,” not for data you may want to release and reclaim.
Walrus: decentralized, priced by use, verifiable
Walrus aims for the practical middle: genuine decentralization and on-chain verifiability, with pricing by size and duration rather than a single perpetual payment. You store for a period, pay in WAL for exactly that, and can renew or release the space. Erasure coding keeps the redundancy cost-efficient — the mechanics are covered in How Walrus Keeps Your Data Alive — and the Sui settlement layer makes availability publicly certified.
- Control: with Walrus, access is bound to your keys — there is no account for anyone to suspend.
- Durability: distributed across independent nodes, recoverable from a subset, certified on-chain.
- Cost model: pay for the space and time you use, not a provider’s margin or a forever-commitment.
- Verifiability: a public blob ID you can audit, rather than a promise on a dashboard.
There is no single “best” storage. There is the right trust model for the data in front of you.
If your priority is deep integration with an existing cloud stack, centralized object storage is hard to beat. If your priority is that no single party can read, lose, price-gouge, or revoke your data — and that its availability is something you can prove — a decentralized network like Walrus is built for exactly that. The simplest way to feel the difference is to store one file yourself: Walrus Drive does it from your own Sui wallet, no account required.