A low-latency, high-throughput orderbook implementation in Rust for trading systems and exchange infrastructure.

This orderbook provides microsecond-level performance for order management and matching with a focus on:

• Memory efficiency
• CPU cache optimization
• Low-latency operations
• Price-time priority matching

The implementation achieves:

• Sub-microsecond latency for core operations
• Throughput exceeding 1.5 million operations per second (M4 Pro 48gb)
• Support for order books with hundreds of thousands of resting orders
• Consistent performance under varied market conditions

• Price-time priority matching engine
• Limit and market order support
• Memory-efficient order representation
• Zero-copy design for critical paths

use orderbook_rs::{OrderBook, Order, Side, OrderType};

// Create an orderbook
let mut book = OrderBook::new("FUEL-USD", 100_000);

// Add a buy order
let buy_order = Order::new(1, 9900, 10, Side::Buy, OrderType::Limit);
let result = book.add_order(buy_order).unwrap();

// Add a sell order
let sell_order = Order::new(2, 10000, 5, Side::Sell, OrderType::Limit);
let result = book.add_order(sell_order).unwrap();

// Check market depth
let (bids, asks) = book.market_depth(10);

// Get orderbook summary
println!("{}", book.summary());

// Cancel an order
book.cancel_order(1).unwrap();

The orderbook uses a Vec-based approach with direct indexing for O(1) price level access, rather than traditional tree-based structures. This design choice prioritizes:

• Contiguous memory layout for better cache locality
• Reduced pointer chasing and memory indirection
• Efficient traversal of price levels during matching

Price levels are constrained to a configurable range around a base price to enable this efficient indexing approach.

Run tests:

cargo test

Run benchmarks:

RUSTFLAGS="-C target-cpu=native" cargo run --release

The long-running benchmark simulates realistic market activity for sustained periods to assess performance stability and throughput.

• Rust 1.85 or higher
• SIMD supported on relevant CPUs

GPL-3.0