Beyond message injection: advanced load testing capabilities for trading systems with th2-shark (th2-loader)

Alyona Bulda, SVP, Emerging Technologies, presented the paper “Beyond message injection: advanced load testing capabilities for trading systems with th2-shark” – co-authored by Iosif Itkin, CEO & Co-founder, Exactpro – as part of the conference program at the 14th International Workshop on Load Testing and Benchmarking of Software Systems (LTB 2026). The one-day workshop took place on 4 May in Florence, Italy and was co-located with the 17th ACM/SPEC International Conference on Performance Engineering (ICPE 2026).

th2-loader (aka th2-shark) is a proprietary Exactpro solution. It is a high-performance load and simulation platform for mission-critical financial market infrastructures. It enables organisations to detect capacity, resilience, correctness, and recovery defects in trading, market data distribution, post-trade, and web-based systems under realistic and extreme conditions, including complex load scenarios.

Keywords
Load testing tool, th2-shark, trading systems, high-load systems, benchmarking, latency analysis, financial protocol testing, performance engineering, production-realistic benchmarking, microburst, microsecond precision load testing, actor model, Golang.

Overview

High-performance trading systems operate in environments where microsecond-level latency differences determine competitive outcomes and where instability under load can be as damaging as functional defects. Validating such systems demands load testing tooling purpose-built for the unique demands of the electronic trading infrastructure capable of sustaining millions of messages per second. It should be able to simulate realistic market participant behavior, support a wide range of financial protocols, and produce actionable, analyst-ready performance reports. This paper presents th2-shark (th2-loader), a high-throughput load generation and benchmarking tool designed specifically for these requirements, and details its architecture, capabilities, integration patterns and application in real-world trading system validation.

Generic load testing tools typically lack support for binary financial protocols, cannot model the stateful, feedback-driven behavior of market participants and introduce measurement overhead that distorts latency results at microsecond resolution. th2-shark was designed to close this gap: a resource-light, protocol-aware, programmable load generator that can faithfully simulate production-like trading conditions while simultaneously capturing the telemetry needed to characterize system performance.

th2-shark is part of Exactpro's th2 next-generation test automation suite. Its core design principle is that test scenarios should be expressive, portable and impose minimal overhead on the system under test.

th2-shark scenarios are implemented in Golang, leveraging the language's concurrency primitives and runtime efficiency for high-throughput workloads. Each scenario compiles to a single portable binary, requiring only a scenario-specific configuration file at runtime. This configuration defines the actor ensemble, connection endpoints, load profiles (steady-state, ramp-up, burst), payload generation rules, and monitoring parameters making th2-shark straightforward to deploy across lab, CI/CD and cloud environments without environment-specific modifications.

The tool's actor model is its primary abstraction for expressing business logic. Actors are programmable entities that can maintain order books at configurable depth levels, subscribe to and arbitrate market data streams, implement feedback-driven trading flows (e.g., reacting to fills with follow-on orders), and simulate the behavior of multiple concurrent market participants. This enables scenarios that go far beyond simple message injection capturing the stateful, interdependent dynamics of real trading sessions.

th2-shark provides comprehensive coverage of the binary protocols used in financial market infrastructures:

  • TCP/IP and SOUP-based: OUCH, SOUP ITCH, GLIMPSE, LSE Native
  • UDP and MoldUDP-based: Mold ITCH, REWIND, MIT ITCH, FAST
  • FIX variants: FIX, FIX SBE (Simple Binary Encoding), FAST

The tool can be deployed against matching engines, smart order routers, market data systems, clearing and settlement infrastructure, and risk management platforms without protocol-specific tooling gaps.

Explore th2-shark (th2-loader)

th2-shark achieves injection rates of hundreds of thousands to millions of messages per second, depending on hardware resources and scenario complexity. Simpler patterns such as rapid new-cancel order cycles approach peak injection rates; production-realistic scenarios involving order management, fills, cancellations and market data processing reduce peak throughput due to increased actor computational demands. The tool supports microburst generation short intra-second load spikes that mimic sudden market activity surges, which is essential for testing system behavior under conditions that steady-state load profiles cannot replicate.

Three analytical layers are built into or tightly integrated with th2-shark (th2-loader):

Real-time monitoring: The tool exposes built-in KPI statistics and supports custom metric emission to Prometheus, enabling correlated dashboards (e.g., message rate vs. CPU/memory/network) during test execution.

Latency analysis: For unencrypted traffic, wire-level timestamps are captured via pcap (Corvil, Keysight, tcpdump). For encrypted communication, latency is derived from timestamped tool logs cross-validated against system traces. Reports are delivered as Excel/CSV with time-series and latency distribution charts and can be ingested into Grafana for multi-dimensional analysis.

Reconciliation: A lightweight post-execution validation layer checks up to 1,000,000 transactions for order book consistency, absence of crosses, gap and duplicate detection and primary/replica flow consistency providing functional confidence without compromising test throughput.

Conclusions

th2-shark addresses a genuine gap in the load testing tooling landscape for financial markets technology. By combining a Golang-based actor model, broad financial protocol support, high injection rates, microburst capability and integrated telemetry. It enables production-realistic benchmarking that generic tools cannot provide.

Alyona Bulda’s slides presented at the LTB 2026 conference are accessible in the Related Materials section of this page.