预测性乐观执行是 Artela 的核心技术之一，也是不同于 Sei 、Monad 等其他并行 EVM 的特点之一。乐观执行指的是一种并行执行策略，假设初始状态下事务之间没有冲突。在这种机制中，每个事务都保持一个私有的状态版本，记录修改但不立即最终确定。事务执行完毕后，进行一次验证阶段，检查是否存在与同时期其他并行事务所引起的全局状态变化的冲突。一旦检测到冲突，就会重新执行事务。预测性是指通过特定的 AI 模型分析历史交易数据，来预测即将执行的交易之间的依赖关系，即哪些交易可能会访问相同的数据，并据此将交易分组安排它们的执行顺序，从而减少执行冲突和重复执行。相比之下，在预测方面，Sei 依赖于开发者提前定义好的交易依赖关系的文件，而 Monad 是采用编译器级别的静态分析生成交易依赖关系的文件，两者都不具备 EVM 等效性，并且都缺乏 Artela 基于 AI 的动态预测模型的自适应能力。
2. 异步预加载技术（Async Preloading）
Asynchronous preloading technology is committed to solving the input and output (I/O) bottlenecks caused by state access, with the purpose of improving data access speed and reducing waiting time during transaction execution. Artela pre-loads the required state data from slow storage (such as hard disk) into fast storage (such as memory) based on predictive models before the transaction is executed. Reduce I/O waiting time during execution by loading necessary data in advance. When data is loaded and cached in advance, multiple processors or execution threads can access the data simultaneously, further increasing execution parallelism.
3. Parallel Storage
With the introduction of parallel execution technology, although transaction processing can be parallelized, if the data reading, writing and updating speed cannot be improved simultaneously, it will limit the overall system performance. The key factor is that the bottleneck of the system gradually shifts to the storage level. Solutions like MonadDB and SeiDB have begun to focus on storage-level optimization. Artela draws on and integrates a variety of mature traditional data processing technologies to develop parallel storage, further improving the efficiency of parallel processing.
The parallel storage system is mainly designed to address two major issues: one is to achieve parallel processing of storage, and the other is to improve the ability to efficiently record data status to the database. During the data storage process, common problems include data expansion during data writing and increased database processing pressure. In order to effectively deal with these problems, Artela adopts the separation strategy of State Commitment (SC) and State Storage (SS). This strategy divides storage tasks into two parts: one part is responsible for fast processing operations and does not retain complex data structures, thereby saving space and reducing data duplication; the other part is responsible for recording all detailed data information. In addition, in order not to affect performance when processing large amounts of data, Artela adopts a method of merging small pieces of data into large pieces, reducing the complexity of data saving.
4. Elastic Block Space (EBS)
Artela’s Elastic Block Space (EBS) is designed based on the concept of elastic computing and can automatically adjust the number of transactions accommodated in a block according to the degree of network congestion.
Elastic computing is a cloud computing service model that allows the system to automatically adjust the configuration of computing resources to adapt to changing load requirements. The main purpose is to optimize resource usage efficiency and ensure that additional computing power is quickly provided when demand increases.
EBS dynamically adjusts block resources according to the specific needs of dApps and provides independent expansion block space for dApps with high demand, aiming to solve the problem of significantly different blockchain performance requirements for different applications. The core advantage of EBS is "predictable performance", that is, the ability to provide predictable TPS for dApps. Therefore, dApps with independent block spaces will receive stable TPS regardless of whether the public block space is crowded or not. In addition, if the contract written by the dApp supports parallelism, it can further achieve higher TPS. It can be said that EBS provides a more stable environment compared to traditional blockchain platforms such as Ethereum and Solana. These traditional platforms often lead to dApp performance degradation when the network is congested, such as during the Inscription boom or during peak DeFi activity. Artela effectively solves such problems through customized and optimized resource management.
In summary, Artela achieves highly scalable and predictable network performance through parallel execution stacks and elastic block space. This parallel execution architecture uses AI models to accurately predict transaction dependencies, reducing conflicts and duplicate executions. In addition, large applications can have dedicated processing capabilities and resources as needed, ensuring stable performance even under high network loads. This enables the Artela network to support more complex application scenarios, such as real-time big data processing and complex financial transactions.
