Facing the limits of fragmented observability

SAS is one of the most trusted names in data and AI, powering decisions that shape healthcare, finance, public safety, and the everyday services people depend on. Governments use SAS to detect fraud and protect citizens. Banks use it to stop real-time financial crime. Pharmaceutical companies use it to accelerate drug discovery. Retailers and logistics providers use it to keep supply chains stable and shelves stocked.

For many SAS engineers, the importance of this work is personal. Principal Performance Engineer Dharmita Lutz recalls supporting a close friend through a difficult cancer treatment decision. “One night I was reading a PubMed article about a treatment option, and at the bottom it said the data had been evaluated using a SAS software package. That was my ‘aha’ moment. I realized that these numbers, these analyses, are what people rely on to make life-changing decisions.” Experiences like hers reflect the responsibility SAS carries as it builds and maintains the systems customers depend on.

At the heart of all this is SAS® Viya®, the company’s cloud-native data and AI platform that processes massive volumes of data and delivers real-time intelligence to organizations under constant pressure to act quickly and accurately. When a decision affects public health, financial stability, or critical infrastructure, customers need Viya to be fast, predictable, and always available.

“Without a unified view, our teams spent too much time piecing together information just to understand where a problem started,” says Joe Flynn, Principal Software Developer at SAS. “We needed visibility that matched the complexity and the mission-critical nature of the platform.”

Gaining clarity through unified observability

SAS chose Datadog to unify the signals their engineers rely on, bringing traces, logs, metrics, database activity, Kubernetes data, and OpenTelemetry spans together into a single source of truth across the company’s distributed architecture. With this consolidated visibility, SAS teams gained the clarity they need to keep insights flowing and to ensure every customer interaction is fast and reliable. “That unified view changed everything,” says Jenn Riemer, Director of Decisioning R&D at SAS.

“Datadog gives our engineers the clarity they need to keep our software fast, dependable, and ready for our customers.”

“With one place to see the full story, we can make decisions with confidence and improve performance at a much faster pace.”

During load tests, SAS engineers can watch real-time traces flow through the platform and pinpoint whether latency stems from downstream services, inefficient queries, CPU contention, or memory spikes. Sharing findings is as simple as sending a link, giving every team the same context and accelerating decision making.

Transforming performance and efficiency

With full system visibility, SAS uncovered new performance opportunities. Engineers identified an N+1 anti-pattern that delivered roughly three hundred times efficiency per call, and they resolved a data access bottleneck that made a key feature seven and a half times faster.

“We've been able to address performance problems with quick, targeted fixes,” says Lutz. “These improvements ripple across the platform, reducing costs and giving customers a noticeably smoother experience.”

These gains produced meaningful infrastructure savings. Using Kubernetes metrics, SAS quantified a four-times reduction in CPU utilization for a critical microservice—resulting in roughly 75% lower compute cost. When multiplied across dozens of services, these optimizations deliver cloud savings and measurable user experience improvements. “It is much easier to pinpoint the root cause of performance slowdowns now,” adds Flynn. “Those insights translate directly into a more responsive platform for our customers.”

Smarter collaboration and faster innovation with AI

SAS has transformed how its engineering teams collaborate by standardizing on Datadog as a shared source of truth. Teams can investigate problems using the same real-time data and visual context. During one cross functional investigation, engineers quickly identified a query optimization, a connection pool bottleneck, and an authorization slowdown in minutes. “It is incredibly easy for our teams to collaborate now,” says Riemer. “Sharing a link gives everyone the same view. That accelerates everything.”

SAS is also advancing its engineering workflows by combining observability insights with AI-assisted development. Once engineers pinpoint where a bottleneck is occurring, AI helps them explore potential root causes and validate improvements more quickly. SAS is evaluating Bits AI, Continuous Profiler, and expanded Real User Monitoring capabilities to extend these workflows across the entire stack.

Building a faster, leaner, more reliable platform

With unified visibility, deeper insight, and rapidly improving performance workflows, SAS has improved both speed and cost efficiency. Teams now operate with shared context, solve issues faster, and uncover new optimization opportunities.

“Datadog helps us build software that is fast, lean, and reliable,” says Riemer.

“Our platform is now more consistent, more predictable, and better aligned with the outcomes our customers expect,” adds Flynn.

*Results shown reflect improvements observed for a single SAS microservice.