For decades, financial market infrastructure has been measured by the familiar metrics of bandwidth, latency and availability. While each is of course important, they no longer tell the whole infrastructure performance story.
As markets become increasingly interconnected, always-on and AI-enabled, financial institutions are discovering that performance is determined less by how fast infrastructure operates under ideal conditions, and more by how predictably it behaves when conditions are not ‘normal’
Speed has long been the defining measure of infrastructure performance. In latency-sensitive trading environments, milliseconds matter. Faster execution, lower latency and greater bandwidth have consistently shaped infrastructure investment, network design and competitive advantage.
While these priorities remain valid, financial markets have evolved. The environment supporting trading, payments and post-trade processing is becoming considerably more complex than today’s operational infrastructure is designed to support.
Always-on payment rails, compressed settlement cycles, AI-driven workloads and distributed market ecosystems require infrastructure to deliver consistent performance under sustained operational pressure, rather than simply meeting (albeit benchmark speeds under controlled conditions.
“Infrastructure should not be judged solely by how quickly it performs under ideal conditions. It should be judged by how predictably it behaves when conditions are anything but ideal.”
Pravin Vijay, Solutions Engineer, Financial Markets
Benchmarks beyond latency
Latency remains fundamental to modern financial markets, and particularly in electronic trading environments where execution quality and market access are linked irrevocably with network performance.
However, latency measured under “laboratory conditions” is only one aspect of infrastructure performance.
Periods of heightened market volatility, geopolitical uncertainty, macroeconomic announcements and unexpected operational events create conditions that place significantly different demands on infrastructure. Artificial intelligence compounds that challenge, generating dynamic traffic patterns that differ markedly from the predictable north-south application flows around which many enterprise networks were originally engineered.
Research from Ciena forecasts a six-fold increase in data centre interconnect bandwidth demand during the next five years, driven largely by AI workloads. These workloads generate increasingly complex east-west traffic between cloud environments, GPU clusters and co-location facilities, creating new patterns of network utilisation that place greater emphasis on routing consistency as well as raw capacity. In these conditions, speed alone becomes an incomplete measure of performance.
Predictability under pressure
Periods of elevated market volatility generate significant increases in market data volumes, trading activity, payment flows and risk calculations. Operational disruptions are relatively few and far between under normal conditions. Infrastructure that performs consistently under routine operating conditions may behave very differently when subjected to sustained and unpredictable demand.
Infrastructure assembled incrementally over years reflects changing business priorities rather than an overarching architectural strategy: Over time, new carriers are introduced. Cloud providers are integrated. Additional routes are added as organisations expand into new markets and onboard new counterparties.
Individually these decisions are entirely rational. Collectively they lead to increasingly complex operating environments in which routing behaviour becomes progressively harder to understand and govern.
This changing environment is reflected in conversations taking place across the financial markets industry which underscore the importance of measuring and managing operational consistency rather than isolated performance metrics (and hoping to recover quickly in the event a disruption occurs).
Performance predictability does not happen by chance – it is the intentional outcome of engineering decisions made throughout the infrastructure stack.
Rather than focusing on how best to connect existing components in the most efficient way, it begins by defining the operational outcomes required before designing the physical network architecture capable of delivering them.
Route diversity, deterministic routing, latency consistency and operational visibility become integral characteristics of the network rather than enhancements introduced later.
Deterministic routing illustrates this distinction particularly well. In many multi-provider environments, traffic may follow different physical paths depending upon network conditions and operational events occurring elsewhere in the infrastructure ecosystem. During routine conditions, these variations may have little observable impact. During periods of market stress, however, even relatively small routing changes can introduce additional uncertainty into operating environments.
Engineering infrastructure around known physical paths reduces that uncertainty, allowing organisations to understand not simply if information reaches its destination, but precisely how it gets there, and how that pattern of behaviour remains consistent under changing operating conditions.
Predictability supports resilience
The increasing emphasis on predictability closely mirrors evolving regulatory expectations. Regulators now expect institutions to understand where operational dependencies exist and how critical services behave under stress.
Frameworks including the Digital Operational Resilience Act (DORA) focus not simply on infrastructure availability but on demonstrating resilience across interconnected ICT environments and critical third-party providers.
Predictability therefore becomes a governance issue as much as a technical one.
Visibility into routing behaviour, operational dependencies and infrastructure performance enables firms to identify emerging issues before they develop into operational incidents. It also supports the evidence increasingly required to demonstrate resilience across complex technology ecosystems.
As cross industry discussions continue to demonstrate, resilience is no longer about recovering effectively from failure; it is about engineering infrastructure intentionally to mitigate the likelihood of failure from the outset.
Redefining infrastructure performance
Financial institutions have spent decades optimising infrastructure for speed.
Speed always matters. Predictability, however, distinguishes infrastructure engineered for today’s financial markets from that built around yesterday’s operating assumptions.
Markets are increasingly interconnected. Instant payments are expected. Settlement windows will continue to compress. Artificial intelligence will fundamentally reshape how data moves across financial networks. Each of these developments increases the importance of infrastructure that behaves predictably and consistently.
The future of financial infrastructure will not be defined by the fastest networks, but by those that display predictable operational outcomes regardless of market conditions. Because in modern financial markets, performance certainty and confidence becomes a competitive advantage.
Principal sources
- Ciena – Global Survey Explores Networking Needs for the AI Era
- European Banking Authority – Operational Resilience
- Regulation (EU) 2022/2554 – Digital Operational Resilience Act (DORA)
- European Central Bank – TARGET Services
- Industry observations synthesised from multiple industry events including FIA International Derivatives Expo (IDX), Quorum 15 and TradeTech Europe.