Mapping vs Array for Dynamic Collection Storage

When designing a Solidity smart contract to manage a dynamic set of user data, the choice of storage structure directly impacts gas consumption and data accessibility. The primary constraint is balancing the need for efficient individual record retrieval against the requirement for on-chain data enumeration.

Mappings offer constant-time lookups but lack native iteration capabilities and a built-in length property. Conversely, dynamic arrays support iteration and tracking of total elements but incur higher gas costs for certain modifications and risk hitting the block gas limit during large-scale loops.

For a system requiring both fast lookups and the ability to list all entries, a common pattern is to implement a mapping paired with an auxiliary array of keys. However, this increases the gas cost for every write operation as two state variables must be updated.

• Does the gas overhead of maintaining a dual-structure (Mapping + Array) outweigh the cost of iterating through a large Array for retrieval?
• At what collection size does the risk of a Denial of Service (DoS) via block gas limit make the Array-only approach non-viable?