Foundations

Core Design Principles

Irys is built around a small set of architectural properties that directly shape how the protocol operates.

This section will give you a high-level view of these properties, while later sections will dive deeper.

• Data persists as network state

  Data written to Irys is intended to remain available for defined durations, including permanent storage. Once promoted to the Publish Ledger, data becomes part of the network's long-term state. Miners are economically responsible for maintaining its availability, and the protocol continuously verifies that it remains retrievable.

• Execution operates on stored data
  Smart contracts can access data that already exists in the network. Execution is designed to interpret, enforce, and coordinate behavior around existing datasets, allowing applications to reference and work with shared onchain data directly within one system.
• Verification is enforced by the protocol

  Availability of stored data on Irys is enforced through:

  • Ingress proofs during promotion
  • Partition-specific replica preparation
  • Random sampling of storage ranges
  • Stake slashing for failure

  Applications rely on these guarantees without implementing their own storage verification mechanisms.

• Storage and execution are managed separately

  Irys separates storage, execution, and programmable data into distinct fee markets. This separation is implemented at the protocol level by processing different transaction categories through dedicated pathways. Storage writes, smart contract execution, and programmable data access are capacity-limited and accounted for independently, so congestion in one pathway does not consume throughput from another.

  Because these components operate independently:

  • Storage pricing can track underlying hardware costs.
  • Execution demand does not increase storage costs.
  • Storage demand does not distort execution fees.

  The protocol therefore does not rely on a single fee market to sustain the network. Storage, execution, and programmable data each contribute independently to network economics.

  This structure allows applications to plan around stable costs and predictable system behavior over time.