TL;DR
This paper introduces a novel blockchain-based framework for fair cloud VM auctions and trades. It addresses the critical issue of trust in cloud auctions by using smart contracts and cryptocurrency to ensure both participants are treated fairly during the bidding and transaction phases, preventing malicious behavior from either the provider or the users.
What Problem Does This Paper Address?
Cloud computing offers flexible and cost-effective resource allocation through virtual machine (VM) auctions. However, existing auction systems often assume a trustworthy auctioneer, which is not always realistic. This assumption can lead to unfairness: malicious users might collude with an auctioneer to gain an advantage, or users might back out of trades after learning bids. Conversely, dishonest providers might fail to deliver services after receiving payment, or dishonest users might receive services without paying. This paper tackles the challenge of ensuring fairness in both the auction and the subsequent trade of cloud VMs, even in the absence of a trusted central authority.
Methods
The core of the proposed solution is a decentralized framework built on blockchain technology, specifically leveraging Ethereum. The authors introduce two key components:
- Smart Contracts: These act as automated, transparent, and tamper-proof auctioneers. They manage the entire auction and trade process, enforcing rules and executing transactions based on predefined logic.
- Cryptocurrency (Ether): Used for financial penalties and rewards. Participants commit "guaranties" (deposits) to the smart contract. Dishonest behavior results in the forfeiture of these guaranties, while honest participation is rewarded.
The framework employs a timed commitment scheme adapted for blockchain. Participants commit to their bids by sending a hash of their bid and a random number to the smart contract. They then reveal their bids within a specified time. Failure to do so incurs a financial penalty.
A crucial element is the state mechanism, which tracks the status of participants and the smart contract. This mechanism ensures that operations are only allowed when specific conditions are met, preventing unauthorized actions and enforcing the protocol.
For the trade phase, two scenarios are considered:
- With an Adjudicator: An independent third party (also managed via the smart contract) verifies the validity of VM instances if disputes arise.
- Without an Adjudicator: A "ladder payment" mechanism is introduced. The VM usage time is divided into segments, and users confirm the use of each segment. This reduces the risk for users by allowing partial payments and partial losses if issues arise, making the trade fairer even without a dedicated dispute resolver.
Datasets
Not discussed in the paper. The evaluation uses simulated networks (Ganache) and the Ethereum Rinkeby test network.
Key Insights and Ideas
- Blockchain as a Trust Anchor: The paper demonstrates how blockchain's inherent transparency and immutability can replace a trusted auctioneer, solving a fundamental problem in cloud resource allocation.
- Financial Incentives for Fairness: Using cryptocurrency for deposits and penalties is a powerful mechanism to enforce honest behavior from all parties. This "skin in the game" approach is key to the framework's fairness guarantees.
- State Machines for Protocol Enforcement: The state mechanism is critical for managing the complex interactions and ensuring that participants adhere to the protocol at each stage, from bidding to trade completion.
- Adaptable Trade Fairness: The introduction of a trade mechanism without an adjudicator, using ladder payments, is a significant contribution. It acknowledges the practical challenges of finding and integrating adjudicators and offers a robust alternative for scenarios where disputes might be resolved through partial losses rather than full arbitration.
Key Results
The paper implements the framework on a local simulated network (Ganache) and the Ethereum Rinkeby test network. Experimental results demonstrate the feasibility of the proposed system.
- Auction Phase Performance: Experiments showed that the auction time is influenced more by the number of users than the number of VM types. For instance, on Ganache, with 5-20 users and 5-9 VM types, the auction time was measured. On the Rinkeby test network, auction times were around 15 seconds, indicating that the auction phase can be chained within a single block time.
- Trade Phase Performance: The ladder payment mechanism, with 5 confirmation steps, also showed consistent timing around 15 seconds on Rinkeby, fitting within block times.
- Fairness Guarantees: The theoretical analysis and proofs demonstrate that the framework achieves both auction fairness (preventing bid manipulation and unfair quitting) and trade fairness (ensuring users get what they pay for and providers get paid for what they deliver), even in the presence of malicious actors.
Limitations and Future Work
- Scalability: While the framework is demonstrated on a test network, the inherent latency and transaction costs of public blockchains like Ethereum could be a limitation for large-scale, high-frequency cloud VM auctions. The paper mentions that Ethereum is unable to afford a lot of computing, leading to choices like limiting the number of users and resources in experiments.
- Complexity of Smart Contracts: Designing and verifying complex smart contracts for such systems is challenging and prone to bugs, which could have significant financial implications.
- Adjudicator Integration: The framework relies on the adjudicator's integrity and availability. The paper acknowledges that finding a suitable adjudicator can be difficult.
- Future Work: The authors suggest further research into optimizing the framework for better scalability and exploring more sophisticated smart contract designs. They also highlight the need for more complex environments and easier implementation.
Summary updated