Delta Storage
The easiest way to save and upload files on IPFS. A hosted IPFS front door built for developers who want the durability guarantees of content-addressed storage without operating IPFS themselves.
Visit siteThe problem
IPFS is powerful but hostile to new users. Pinning strategies, gateway performance, CID variants, garbage collection, and reliability engineering are concerns most product teams don't want to solve. The alternatives — running your own IPFS node cluster or piecing together third-party pinning services — trade one set of ops burdens for another. A friendly IPFS front door — simple API, transparent pricing, reliable pinning — is table stakes for consumer adoption of decentralized storage. Building it well means solving the operational problems once, at platform scale, so downstream teams don't have to.
Key challenges
Hosted IPFS at scale has all the ops problems of consumer file storage (auth, quota, abuse, cost) plus IPFS-specific problems (pinning strategy, gateway performance, garbage collection, retrieval reliability from the broader network). The service has to be reliable enough to be a dependency for other products (like Koneksi) while staying affordable enough to be attractive against centralized alternatives.
What we built
AR Data built Delta Storage — a hosted IPFS upload service with a simple API, transparent pricing, and reliable pinning. The service handles pinning across a managed node cluster, gateway serving with performance SLAs, garbage collection tuned per tenant, and the operational surface that lets developers use IPFS without operating IPFS. Every upload produces a CID with predictable pinning duration and retrieval performance. The API surface is intentionally small: upload, retrieve, delete. Pricing is per-GB per-month with no per-request fees, which matches what developers expect from consumer file storage.
Our approach
- 1
Small API surface, sensible defaults
IPFS complexity is not the user's problem. The API is deliberately minimal — upload, retrieve, delete — with sensible defaults for pinning strategy and CID versions.
- 2
Managed node cluster with performance SLAs
Running IPFS well is real ops work. Managing that once at platform scale means downstream consumers don't have to.
- 3
Tenant-tuned garbage collection
GC strategy has real cost implications. Tuning it per tenant based on access patterns keeps performance and cost balanced.
- 4
Simple, transparent pricing
Per-GB per-month with no per-request fees. Developers can predict cost; that predictability is what makes the service adoptable.
Key architectural decisions
Small API surface over configurability
Configurability is complexity. A small API with good defaults is what makes hosted IPFS actually usable for non-specialists.
Managed cluster over per-tenant deployments
Per-tenant IPFS clusters have prohibitive ops cost. A managed shared cluster with logical isolation is the tractable model.
Per-tenant GC tuning
One-size-fits-all GC leaves cost and performance both suboptimal. Per-tenant tuning fits the actual access patterns.
Predictable pricing over metered complexity
Developer adoption favors predictable cost. Metered per-request pricing creates cognitive load that reduces adoption.
Results
- Simple upload API with sensible defaults
- Reliable pinning infrastructure with performance SLAs
- Transparent per-GB pricing without per-request complexity
- Developer-first UX — hosted IPFS without operating IPFS
- Runs as dependency infrastructure for other AR Data products (Koneksi)
- Managed cluster with per-tenant GC tuning
Impact
Delta Storage is the operational foundation that lets other decentralized-storage products (like Koneksi's ransomware protection) exist without each of them re-solving IPFS operations. The engagement is why decentralized storage remains a first-class service line — we're not asking clients to trust us with a research project; we run the infrastructure ourselves.
Tech stack
Want a case study like this?
30 minutes. We scope the real problem and figure out what to build.
Book a call
