|
| 1 | +# Strata design rationale |
| 2 | + |
| 3 | +In designing Strata, |
| 4 | +we adopted a structure inspired by Ethereum's post-merge (formerly "2.0") architecture, |
| 5 | +where each full node consists of two distinct components: |
| 6 | +an **execution client** and a **consensus client**. |
| 7 | +In Ethereum, the consensus client manages Proof of Stake (PoS) consensus, |
| 8 | +including proposer selection and block finalization, |
| 9 | +while the execution client is responsible for transaction execution using |
| 10 | +the EVM interpreter and processing user-facing application-level queries. |
| 11 | +These two clients communicate via the **Engine API**, |
| 12 | +which allows them to exchange necessary information, |
| 13 | +such as when the consensus client requests |
| 14 | +the execution client to process a block. |
| 15 | + |
| 16 | +Similarly, the Strata network is divided into two blockchains, |
| 17 | +with each full node able to run a client for each of the following layers: |
| 18 | + |
| 19 | +1. **Execution layer**: Handles transaction execution and chain state. |
| 20 | +1. **Orchestration layer** |
| 21 | + (our version of Ethereum's consensus layer): |
| 22 | + Coordinates various specialized tasks related to bitcoin interactions, |
| 23 | + bridge operations, decentralized sequencing, |
| 24 | + and other Strata-specific operations. |
| 25 | + |
| 26 | +In our architecture, the execution layer is an EVM environment implemented |
| 27 | +using a fork of Reth, |
| 28 | +while the orchestration layer is a new blockchain tailored to implement |
| 29 | +Strata's rollup design and support its consensus requirements. |
| 30 | +The two layers communicate via a slightly modified version of the Engine API, |
| 31 | +which we extended to handle specific features like bridge transfers in/out of |
| 32 | +the execution layer. |
| 33 | +This modification was necessary as Ethereum's Engine API doesn't natively support |
| 34 | +this functionality. |
| 35 | + |
| 36 | +## Advantages of separation |
| 37 | + |
| 38 | +By following Ethereum's well-thought-out approach to separating execution and orchestration, |
| 39 | +Strata gains important advantages: |
| 40 | + |
| 41 | +- **Modularity and specialization**: |
| 42 | + The execution and orchestration layers have distinct, specialized functions. |
| 43 | + This separation allows them to be developed, maintained, |
| 44 | + and iterated upon independently. |
| 45 | + For example, improvements to the orchestration layer's coordination logic |
| 46 | + can be made without impacting the execution layer, and vice versa. |
| 47 | + |
| 48 | +- **Future support for multiple execution environments**: |
| 49 | + This separation allows us to support different execution environments in the future, |
| 50 | + each conforming to our modified Engine API. |
| 51 | + This means a future with different kinds of execution layers |
| 52 | + (eg. WASM-based, privacy-focused, etc.), on top of the orchestration layer, |
| 53 | + all having the ability to easily and trustlessly interoperate. |
| 54 | + |
| 55 | +## Why we chose a custom solution over existing SDKs |
| 56 | + |
| 57 | +We initially started development with an existing SDK but, |
| 58 | +after several months, chose to start over from scratch due to |
| 59 | +the challenges of relying on third-party tools. |
| 60 | +While SDKs like **Sovereign SDK**, **Rollkit**, |
| 61 | +and **OP Stack** could have sped up early development, |
| 62 | +we opted to build a custom orchestration layer for several key reasons: |
| 63 | + |
| 64 | +- **Complete control and flexibility**: |
| 65 | + Our main objective was to retain full control over the development of |
| 66 | + our orchestration and consensus logic. |
| 67 | + This allows us to introduce advanced features—such as instant preconfirmations, |
| 68 | + slashing for misbehavior, interoperability, |
| 69 | + and volition-style hybrid DA—on our own timeline, |
| 70 | + without being restricted by external SDK limitations. |
| 71 | + While pre-built SDKs provide basic functionalities, |
| 72 | + extending beyond those often requires coordination with external teams, |
| 73 | + which may create friction if the roadmaps are misaligned. |
| 74 | + Building our own solution ensures we can evolve our architecture independently. |
| 75 | +- **Avoiding development bottlenecks**: |
| 76 | + During our initial development with a third-party SDK, |
| 77 | + we encountered multiple limitations that slowed our progress. |
| 78 | + In one instance, something as simple as retrieving |
| 79 | + the status of an EVM transaction on the bitcoin chain required suboptimal workarounds |
| 80 | + due to the difficulty in modifying the SDK's interface. |
| 81 | + Although the SDK team eventually supported us |
| 82 | + in making SDK changes or providing better solutions, |
| 83 | + relying on external fixes introduced delays. |
| 84 | + By building our orchestration layer from scratch, |
| 85 | + we eliminate these bottlenecks, |
| 86 | + enabling faster iterations and more efficient development. |
| 87 | +- **Limitations of current SDKs for our use case**: |
| 88 | + Most of the available third-party SDKs, mostly by design, |
| 89 | + lack native support for rollups with bitcoin as the settlement layer. |
| 90 | + OP Stack, for instance, is centered on optimistic rollups |
| 91 | + and fault proofs on Ethereum. Sovereign SDK and Rollkit, while flexible, |
| 92 | + are built around generic rollup use cases and typically integrate with EVM |
| 93 | + or Cosmos-like ecosystems. |
| 94 | + In contrast, Strata's architecture is specifically designed for a rollup |
| 95 | + that settles to bitcoin and includes a validating bridge to and from bitcoin. |
| 96 | + This requires unique design considerations, |
| 97 | + such as accommodating bitcoin's UTXO transaction model, |
| 98 | + limited scripting capabilities, and probabilistic settlement. |
| 99 | + Our bitcoin-centric approach enabled us to easily incorporate a |
| 100 | + BitVM-based bridge from the outset and provides flexibility to |
| 101 | + integrate future bitcoin-specific innovations more easily. |
| 102 | + |
| 103 | +--- |
| 104 | + |
| 105 | +By building our orchestration layer from scratch, |
| 106 | +we not only gain complete control over how we manage consensus, |
| 107 | +bridge operations, and interactions with bitcoin, |
| 108 | +but we also position Strata for future extensibility without being locked into |
| 109 | +the limitations of existing SDKs or external dependencies. |
| 110 | +We do learn from the existing SDKs and borrow ideas where we can, |
| 111 | +but our custom solution allows us flexibility and specificity for our use case. |
0 commit comments