Periodic Reporting for period 1 - STARCloud (Bringing Quality of Service to Blockchain-Based Clouds)
Periodo di rendicontazione: 2019-06-10 al 2021-06-09
Cloud computing has revolutionised computation by facilitating resource pooling and reducing costs. Today the cloud market, however, is restricted to few providers, which entails higher prices, vendor lock-in and lower service quality. Smart Contracts (SCs) and blockchain have the potential to reshape this market. SCs are agreements described in executable language to enable trusted transactions. Blockchain securely registers transactions, even in completely decentralised systems. Together, they bring predictable results to clouds without resorting to intermediaries, forming the so called blockchain-based clouds. These clouds could make market's barriers so low that anyone with idle or spare computational resources could offer cloud services and consumers could easily change provider. However, this paradigm currently adopts the best-effort model, which does not provide guarantees to consumers, not even in case of low performance or failures. Supporting Quality of Service (QoS), that is, defining and enforcing terms, conditions and penalties for the different dimensions of the service provision, such as, availability and response time, is essential; but the state-of-the-art solutions of other paradigms, e.g. traditional clouds, cannot be used in this context.
In the STARCloud project we extended the SLAC SLA language to cover the most important aspects of cloud computing and blockchain-based clouds. The developed SLAC extension is flexible and permits the definition of cloud SLAs, including also the most important features of Smart Contracts, such as, the conditions and the changes of contract terms during runtime.
The project also defined the main phases and mechanisms of the service life cycle in blockchain-based clouds and developed solutions for some of these phases. To validating service runtime monitoring data and executing smart contracts, we developed and evaluated a framework, namely COaaS (Crowd-based Oracle as a Service). It enables the creation of decentralised oracle solutions based on a customizable number of oracles that do reach consensus on the outcomes of QoS monitoring. It aims to overcome some of the main limitations in the interaction between the real-world and the blockchain: (a) the need for trusted third-parties; (b) the lack of transparency; (c) a single-point of failure in case of exploiting a single oracle. In our framework, a blockchain is leveraged to improve security in the process of reaching a consensus among distributed oracles.
Together with iExec, which contributed with data for validation and a another perspective to the discussions, we also defined a model to predict provider reliability, that is, the probability of failure-free execution of computational tasks and correctness of the computed outputs, by extracting the potential dependencies between providers from historical log traces. This model can then be utilized in the definition of provider reputation or in the scheduling of new services. We also proposed a probabilistic scheduler that chooses the providers that meet the reliability constraints.
The project also defined the main phases and mechanisms of the service life cycle in blockchain-based clouds and developed solutions for some of these phases. To validating service runtime monitoring data and executing smart contracts, we developed and evaluated a framework, namely COaaS (Crowd-based Oracle as a Service). It enables the creation of decentralised oracle solutions based on a customizable number of oracles that do reach consensus on the outcomes of QoS monitoring. It aims to overcome some of the main limitations in the interaction between the real-world and the blockchain: (a) the need for trusted third-parties; (b) the lack of transparency; (c) a single-point of failure in case of exploiting a single oracle. In our framework, a blockchain is leveraged to improve security in the process of reaching a consensus among distributed oracles.
Together with iExec, which contributed with data for validation and a another perspective to the discussions, we also defined a model to predict provider reliability, that is, the probability of failure-free execution of computational tasks and correctness of the computed outputs, by extracting the potential dependencies between providers from historical log traces. This model can then be utilized in the definition of provider reputation or in the scheduling of new services. We also proposed a probabilistic scheduler that chooses the providers that meet the reliability constraints.
The STARCloud project will provide novel methodologies and tools for supporting QoS in the next generation clouds, which are particularly challenging due to their decentralised nature. Entering in the traditional cloud market and competing with cloud giants (Google, Amazon, Microsoft, etc.) is extremely difficult but blockchain creates an opportunity for European companies to become more competitive since it enables open markets. The solutions proposed here will contribute to this disruption, having immediate use in many different contexts, by addressing the lack of a requirement of most consumers, in particular business: QoS definition and enforcement. More specifically, it will provide novel approaches and tools for blockchain-based clouds, including consensus mechanisms, a DSL and a framework for the matchmaking and negotiation of SCs, which address important research questions, such as, how to enforce QoS without trusted intermediaries and how to provide more flexibility to QoS definition by using the expression power of SCs; thus overcoming one of the main limitations of blockchain-based clouds, and the lack of support for QoS in decentralised environments.