Technical overview

Architecture

Indexing without Icedancer

• Every project requiring indexing runs it's own archival solana node, a custom Geyser plugin, a database and an API gateway.

• This results in duplication of effort and high cost.

• Given the high speed and size of Solana transactions, the hardware requirement is high.

Indexing with Icedancer

• RPC, geyser plugin, database and API gateway is pooled. This results in significant cost savings.

• Each project deploys its subgraph to Icedancer, which is executed using Wasmer. Transactions having the subgraph's linked account are passed to the WASM module.

• The WASM module processes the input. It has the ability entities to database.

• Database entities can be consumed via a GraphQL API.

Tech stack

• Forked and optimized Postgres geyser plugin

• Define entities and API in GraphQL

• Subgraphs written in Rust and compiled to WASM

• Indexing server written in Rust that executes subgraphs with Wasmer

• Postgres database for storage