As the efficiency and speed gains from this technology become increasingly evident with respect to traditional serial execution engines like the Ethereum Virtual Machine (EVM), parallelism is rapidly transitioning from a strategic advantage to an industry standard.
Under the Hood: Parallelism
The evolution towards parallelised execution layers is not a passing industry exploration; it's a response to a longstanding challenge in blockchain technology. Maintaining a correct, logical order for the execution of every transaction in a block is crucial. Without this order, consensus breaks, and logic fails. Validators, after proposing and agreeing on block orders, face the challenge of executing transactions in a way that ensures all arrive at the same final state. To address this challenge, block proposers and validators have traditionally verified transactions sequentially, ensuring there are no dependencies. This sequential verification contributes to a bottleneck, resulting in slow transaction times as ordering correctly takes time in traditional blockchain designs.
Ecosystem developers responded by introducing Block-STM—a highly efficient, multi-threaded, in-memory parallel execution engine. Block-STM introduces an ingenious parallel execution engine that leverages Software Transactional Memory techniques with optimistic concurrency control. It adopts a pipelined and modular approach for the key stages of transaction processing that allow Block-STM to make late-stage execution decisions around gas, parallelism, and other objectives deterministically. This groundbreaking approach ensures that all transactions are eventually committed while avoiding the pitfalls of traditional STM libraries.
Industry Traction and Adoption
The implementation of Block-STM has set the standard for safe and simple parallel execution, igniting adoption in our ecosystem—and beyond—to transform the industry as we know it. For example, Polygon co-founder Sandeep Nailwal believes, "More and more parallelisation is coming faster than we all think," affirming the widespread adoption of parallel execution. It’s amazing to see peers like Polygon building in public to bring Block-STM to EVM. By carrying on the torch, parallelization is working its magic on EVM.
Members of the Aptos ecosystem, like Econia, also emphasize that their on-chain order book would not have been possible on any other network. Specifically, step functions like parallel execution, optimistic concurrency and dynamic memory allocation have allowed Econia to usher a new era of open finance on Aptos—at scale and without the tradeoffs or limitations typically present on serial execution engines.
The value of parallelism extends far beyond the Aptos ecosystem. For example, chains like Sei have implemented solutions based on Block-STM’s design that aim to improve concurrent execution performance. Monad similarly has validated the STM-based optimistic concurrency approach over different execution models, such as memory locks and actor models. As the performance gains from this technology become increasingly evident, Aptos Foundation is proud of the innovation first forged in the Aptos ecosystem. The growing adoption of parallel execution as an industry standard to date is testament to its merit and potential for Web3.
Block-STM's adaptability to varying workloads and its seamless integration into the Aptos blockchain position it as a cornerstone in blockchain design. Beyond being a parallel execution engine; it's a testament to Aptos ecosystem contributors' commitment to pushing the boundaries of Web3 technology. At the Aptos Foundation, we’re excited that the Aptos ecosystem has been able to leave mark after mark on blockchain design.
As Aptos Network is at the cutting edge of open-source innovation, the industry can anticipate continued breakthroughs in scalability and performance. The collaboration and community-driven ethos of Aptos contributors sets the stage for a pivotal 2024 in the continued development of parallelism.
Click here to learn more about BlockSTM by the Aptos ecosystem.
