Gas Refunds and SSTORE Optimization on Ethereum

Definition

On the Ethereum blockchain, gas refunds referred to a mechanism that partially reimbursed users for gas fees when their transactions reduced the network's state size. This primarily occurred when smart contracts cleared storage slots or were destroyed using the SELFDESTRUCT opcode. The SSTORE opcode, fundamental to Ethereum, is responsible for writing data to a contract's storage, and its gas cost varies significantly based on whether it's setting a value, changing an existing one, or clearing a slot. These mechanisms were designed to encourage efficient state management and prevent the blockchain from becoming excessively large.

Gas Refund: A partial reimbursement of transaction gas fees, historically granted on the Ethereum network when a transaction reduced the blockchain's state size, typically by clearing storage slots or self-destructing a contract.

SSTORE: An Ethereum Virtual Machine (EVM) opcode used by smart contracts to write or modify data in their persistent storage. The gas cost associated with SSTORE operations is dynamic and depends on the nature of the storage modification, such as writing a new value, updating an existing one, or clearing a slot.

Key Takeaway

Gas refunds were initially designed to incentivize efficient use of blockchain storage by making it cheaper to free up space. However, this mechanism was exploited by "gas tokens," which led to inefficient use of block gas limits. Consequently, most gas refunds were removed by EIP-3298 to enhance network stability and reduce complexity. Despite this change, SSTORE optimization remains a critical aspect of smart contract design, as developers continue to strive for minimizing gas costs for users by efficiently managing storage and avoiding unnecessary writes.

Mechanics

The operation of gas refunds and the costs associated with the SSTORE opcode are intrinsically linked to how Ethereum manages its blockchain state. Every smart contract on Ethereum possesses persistent storage, composed of 256-bit words. Writing to this storage is one of the most expensive operations within the Ethereum Virtual Machine (EVM). The gas costs for the SSTORE opcode have been adjusted over time through various Ethereum Improvement Proposals (EIPs), notably EIP-2200 (Istanbul) and EIP-2929 (Berlin), to refine incentives for state management and improve network efficiency.

Prior to EIP-3298, gas refunds were granted when a storage slot was set from a non-zero value to zero (i.e., cleared) or when a contract was entirely removed using SELFDESTRUCT. The objective was to motivate developers to remove unnecessary data from the state, thereby controlling the blockchain's size. The costs for SSTORE operations varied significantly: writing from zero to a non-zero value cost 20,000 gas, while changing a non-zero value to another non-zero value cost 5,000 gas. Setting a non-zero value to zero also cost 5,000 gas but additionally generated a refund. These refunds were capped at a maximum of 50% of the total transaction gas costs to prevent abuse, but this limit proved insufficient to curb all exploits. The dynamic nature of these costs and refunds added a layer of complexity to gas estimation and contract optimization.

Trading Relevance

For traders and users interacting with decentralized applications (dApps) and smart contracts, understanding gas refunds and SSTORE optimization holds indirect but significant importance. Although gas refunds were not directly visible to the end-user, they influenced the overall cost of transactions and, consequently, the economic viability of interacting with certain protocols. Developers who designed their smart contracts efficiently and optimized SSTORE operations could enable lower gas fees for their users, which in turn increased the attractiveness and usability of their dApps. This was particularly relevant during periods of high network congestion, where every unit of gas counted.

The removal of gas refunds by EIP-3298 has fundamentally altered the landscape of smart contract optimization. Developers can no longer rely on refunds to reduce the cost of freeing up storage. Instead, they must design contracts from the outset that utilize storage as efficiently as possible and avoid unnecessary writes. For traders, this means that gas fees for certain operations that previously benefited from refunds may now be potentially higher. It underscores the necessity of closely monitoring gas strategies when interacting with complex DeFi protocols or minting NFTs, as the efficiency of the underlying contract directly impacts transaction costs and overall user experience.

Risks

The mechanics of gas refunds, particularly before their removal, posed certain risks and challenges for the Ethereum network. The most prominent risk was the potential for economic exploits through so-called "gas tokens." These tokens exploited the refund mechanisms by "storing" gas when gas prices were low (by writing data to storage) and then "releasing" it when gas prices were high (by clearing the data to receive refunds). This allowed users to smooth out effective gas costs and potentially manipulate block gas limits, leading to inefficient use of block capacity and impacting network stability.

Another risk arising from the complexity of gas refunds was the difficulty in optimally designing smart contracts. Faulty implementation of storage optimizations could lead to unexpectedly high gas costs or even contract errors. With the removal of refunds by EIP-3298, these specific exploits are no longer possible, but the need for careful SSTORE optimization persists. Developers must now be even more proactive in minimizing state size and ensuring efficient storage access, as there is no longer a "second chance" through refunds. Failure to do so can result in higher fees for users and reduced adoption of the dApp, highlighting the ongoing importance of robust contract engineering.

History and Examples

The history of gas refunds and SSTORE optimization is closely intertwined with the evolution of the Ethereum network and its efforts to scale and improve efficiency. Initially, refunds were introduced to create an incentive for state reduction, as clearing data from storage was expensive, and the network tended to become bloated with unnecessary data. EIP-2200, implemented in the Istanbul hard fork (2019), marked a turning point. It introduced "Net Gas Metering" for SSTORE, redefining gas costs based on changes to the stored value and specifying refunds for clearing storage slots. For instance, writing from zero to a non-zero value cost 20,000 gas, while changing a non-zero value to zero cost 5,000 gas and generated a refund.

A prominent example of the exploitation of this mechanism was the emergence of gas tokens such as GST2 and CHI. These smart contracts allowed users to "tokenize" gas by writing data to storage when gas prices were low. Later, when gas prices were high, this data could be cleared to receive gas refunds, which were then used to fund other transactions. This led to an artificial increase in block gas limits and inefficient use of network resources, as blocks were filled with "empty" transactions solely aimed at obtaining gas refunds. In response to these exploits and associated concerns about state bloat, EIP-3298 was implemented in the London hard fork (2021), removing most gas refunds for SSTORE and SELFDESTRUCT. This marked a paradigm shift, moving the focus from incentives for state reduction to a more direct and robust cost structure to ensure network stability and predictability.

Common Misunderstandings

A widespread misunderstanding is that gas refunds on Ethereum are still fully active. Many users and even some developers are not fully aware of the profound changes brought about by EIP-3298. The reality is that most refund mechanisms, particularly for SSTORE operations and SELFDESTRUCT, have been removed. This means that incentives for state reduction through direct cost reduction no longer exist. Instead, developers must integrate efficient storage management into their smart contracts from the outset, as clearing data still costs gas but no longer generates a significant refund. This misunderstanding can lead to incorrect assumptions about the gas costs of transactions and inefficient contract interactions.

Another common misconception is that the SSTORE operation is always extremely expensive, regardless of context. While the initial write from zero to a non-zero value (20,000 gas) is indeed very costly, subsequent changes to an already occupied storage slot (non-zero to non-zero) are significantly cheaper at 5,000 gas. Furthermore, within the same transaction, gas costs for SSTORE operations after the first modification are even lower, as the value is already present in the EVM's memory. It is also a fallacy that optimizing gas fees is solely a developer's task. While developers bear primary responsibility, end-users directly benefit from well-optimized contracts through lower transaction costs. Understanding these nuances is crucial for accurately assessing the true costs and efficiency of smart contract interactions.

Summary

Gas refunds and the optimization of the SSTORE opcode were, and continue to be, central themes for the efficiency and scalability of the Ethereum blockchain. Originally conceived as an incentive to reduce state size, gas refunds allowed for a partial reimbursement of fees when storage space was freed up. This mechanism, especially after EIP-2200, aimed to curb state bloat and improve network performance. However, the introduction of gas tokens, which exploited this mechanism, led to undesirable side effects, such as the artificial inflation of block gas limits and inefficient resource utilization.

The Ethereum network's response to these challenges was the implementation of EIP-3298, which removed most gas refunds. This step marked a significant shift in Ethereum's design philosophy, moving away from complex incentive structures towards a simpler and more predictable cost basis. Although direct refunds are now largely a thing of the past, SSTORE optimization remains of paramount importance. Developers must continue to design smart contracts that efficiently utilize storage and minimize unnecessary writes to keep transaction costs low for users. For anyone interacting with Ethereum, understanding this evolution and the current cost structure is crucial for making informed decisions and anticipating the impact on fees.