Modern applications increasingly rely on specialized databases and platforms to power real-time analytics and support advanced AI/ML capabilities. These tools help teams accelerate development by consolidating workflows and processes, enabling faster and more efficient data operations. That’s why Datadog has launched three new data platform integrations with Supabase, DuckDB, and Milvus. These integrations give users the visibility they need to monitor their data platform performance, diagnose issues efficiently, and optimize resource allocation across their infrastructure.
In this post, we’ll explore how each new Datadog integration helps users monitor their data infrastructure, including:
- Serverless data workflows with Supabase
- In-process analytics with DuckDB
- Vector search for AI/ML with Milvus
Serverless data workflows with Supabase
Supabase brings together a serverless PostgreSQL database, API endpoints, authentication, and storage into a single platform, enabling developers to build rich applications quickly. Because Supabase handles so many underlying functions, having robust observability is crucial to understanding system behavior, diagnosing performance bottlenecks, and optimizing resource utilization.
Datadog’s Supabase integration gathers data from each Supabase component and correlates it in one place. The integration provides users with real-time insights into query behavior, request patterns, and infrastructure usage for faster diagnosis of latency spikes and resource contention. By providing a unified view of these signals, Datadog helps developers build new features faster, shorten incident response cycles, and gain tighter control of resource costs.
In-process analytics with DuckDB
DuckDB is an in-process analytical database that excels at running SQL queries over local files, making it a popular choice for data exploration and embedded analytics. Datadog’s DuckDB integration tracks ongoing threads, memory usage patterns, and write-ahead log activity, enabling teams to maximize performance and resource efficiency. Users can correlate these metrics with broader infrastructure events to help identify causes of unexpected slowdowns or out-of-memory errors.
With Datadog and DuckDB, organizations can alert on surges in the number of worker threads that would lead to increased CPU usage. Organizations can also configure alerting to flag anomalies in other DuckDB performance telemetry data early, protecting critical jobs from bottlenecks. This unified visibility empowers data teams to optimize how they partition datasets, schedule resource-intensive jobs, and manage concurrency levels so that DuckDB operates at peak efficiency.
Vector search for AI/ML with Milvus
Milvus is an open source vector database built for generative AI workloads, such as semantic search, recommendation systems, and other search queries that require very low latency. As data volumes grow and query rates increase, small inefficiencies in indexing, caching, or resource utilization can ripple into major performance issues.
Datadog’s Milvus integration provides insight into indexing timelines, cache hit rates, and query latencies, delivering a comprehensive view of how Milvus performs alongside the rest of an AI stack. By monitoring indexing performance in tandem with model-serving components (for example, TorchServe, NVIDIA Triton Inference Server, and vLLM), teams can troubleshoot spikes in latency or memory usage before the spikes impact user-facing features. This integration helps teams analyze trends in Milvus operations and make informed scaling decisions. As a result, vector-based search can remain efficient and responsive, regardless of data volume or query rates.
Monitor your modern data platforms with Datadog
From serverless architectures to embedded analytics and AI/ML vector search, Datadog’s integrations for Supabase, DuckDB, and Milvus provide visibility into the performance and health of each technology. By consolidating metrics, logs, and traces in one platform, Datadog helps teams reduce the complexity of monitoring modern data stacks, mitigate risks before they impact end users, and deliver high-performing applications.
You can explore the Datadog documentation for Supabase, DuckDB, and Milvus to learn how to get started. To find out more about how Datadog helps you monitor each layer of your AI stack—including infrastructure, data management, model serving and deployment, foundation models, and service chains and applications—read our AI integration roundup blog post.
If you’re new to Datadog, sign up for a 14-day free trial today.
