At Datadog, over 2,000 engineers deploy and ship new features daily. As a leading observability and security platform used by thousands of companies, ensuring quality and reliability is no small feat. Part of our commitment to excellence lies in our dogfooding culture where our engineering organization is one of the largest and most de...

Datadog operates dozens of Kubernetes clusters, tens of thousands of hosts, and millions of containers across a multi-cloud environment, spanning AWS, Azure, and Google Cloud. With over 2,000 engineers, we needed to ensure that every developer and application could securely and efficiently access resources across these various cloud pr...

Temporal is an open source platform to build resilient and reliable distributed systems. Datadog started using Temporal in 2020 as the foundation for our internal software delivery platform. Since then, its usage has been widely adopted as a platform that any engineering team can use to build their systems.

OpenTelemetry (OTel) is an open source, vendor-neutral observability framework that supplies APIs, SDKs, and tools to instrument, generate, collect, and export telemetry data (metrics, logs, traces and soon profiles). It has a vibrant ecosystem of components, integrations and vendors.

In modern dynamic environments there is a requirement to be able to quickly respond to signals and change the configuration of your applications and systems. As Datadog supports an ever widening set of customer use cases, it has become necessary to allow for updates to systems directly from Datadog's platform.

As companies rapidly adopt Large Language Models (LLMs), understanding their unique challenges becomes crucial. Join us for a special episode of 'Datadog On LLMs: From Chatbots to Autonomous Agents,' streaming live from DASH 2024 on Wednesday, June 26th, to discuss this important topic.

Container orchestration platforms, like Kubernetes, were, from the beginning, an ideal solution for microservice architectures running a lot of stateless services. This was also the case for Datadog, which is run on dozens of self-managed Kubernetes clusters in a multi-cloud environment, adding up to hundreds of thousands of pods. But ...

In this episode we'll visit the world of predictive analytics and machine learning and uncover how these cutting-edge technologies are transforming the way Datadog monitors and improves its services.

Datadog, the observability platform used by thousands of companies, runs on dozens of self-managed Kubernetes clusters in a multi-cloud environment, adding up to tens of thousands of nodes, or hundreds of thousands of pods. Also, this infrastructure is used by a wide variety of engineering teams at Datadog, with different features and ...

Over the last five years, the Datadog platform has grown. We added Application Performance Monitoring to complement our core infrastructure monitoring product, Log Management, Synthetic and Real User Monitoring, and more. For an enterprise software platform to be successful, the whole has to be greater than the sum of its parts. In Dat...

For many engineers, Identity Management can elicit a broad range of emotions—from confusion during setup and configuration, to complete disinterest as it disappears into the background during day-to-day work, to frustration they encounter erroneously blocked access, and sometimes to terror when misconfigurations lead to a breach. In th...

Datadog, the observability platform used by thousands of companies, runs on dozens of self-managed Kubernetes clusters in a multi-cloud environment, adding up to tens of thousands of nodes, or hundreds of thousands of pods. This infrastructure is used by a wide variety of engineering teams at Datadog, with different feature and capacit...

The extended Berkeley Packet Filter, eBPF has resulted in an ecosystem of new tooling that allows running programs in the linux kernel without loading kernel modules. eBPF seeks to do this safety within a secure sandbox environment and has been a boon to observability and security.

Understanding the health and user experience of your mobile application is critical in order to avoid user frustration, understand application crashes, and reduce bugs mean time to resolution. To help with that task, Datadog has a mobile monitoring solution that allows developers to better understand and improve their application. But ...

WebRTC is a standard for real-time digital communications by enabling video, audio, and data streaming. Originally created for web browsers, Datadog uses WebRTC to create streaming applications across a variety of platforms, from Electron-based native applications to mobile applications.

Caching (and cache invalidation!) is often mentioned as one of the hardest problems in computer science. While caching can bring substantial performance improvements, reasoning about cached data can be extremely difficult as caching fundamentally means that you are no longer reading from your source of truth. With that in mind, many te...

Datadog is an observability and security platform that ingests and processes tens of trillions of data points per day, coming from more than 22,000 customers. Processing that amount of data in a reasonable time stretches the limits of well known data engines like Apache Spark.

There are many different ways to implement Site Reliability Engineering (SRE). From team structures to roles and responsibilities to planning and prioritization flows, there’s no golden path for how to organize things. As Datadog has shifted from a startup to a quickly-growing public company, we’ve seen our own SRE practice evolve. Wit...

The security industry is full of complex terminology like threat, vulnerability, and mitigations. Definitions matter as we design processes that scale. At Datadog, the Security Research functions are focused on detection and response to specific types of threats and vulnerabilities. Workload vulnerabilities, cloud control plane vuln...

When Datadog introduced its Log Management product, it required a new event data storage platform, as storing logs and events is a completely different problem from storing metrics, which was the first Datadog product.

Datadog, the observability platform used by thousands of companies, is made up of hundreds of services that communicate over the network using gRPC, an RPC framework, making it a critical component for Datadog’s reliability.

Datadog is an observability and security platform. That means that our users may be in a high stress situation: debugging an issue in production, managing an incident or responding to a security threat. Having a good UX is particularly critical in those cases.

At Datadog’s scale, with over 18,000 customers sending trillions of data points per day, analyzing the volume of data coming in can be challenging. One of the largest log sources internally at Datadog are networking logs. Being able to analyze and make sense of them is critical to keep Datadog secure. To help with the task, we have bui...

Modern web applications are incredibly complex. Frameworks, javascript, and dependency management have made understanding and maintaining a baseline security standard maximum difficulty. With attack vectors like those listed in the OWASP Top 10 it can be incredibly difficult to know where to start and what the metrics for success are. ...

Rust is a programming language that has been gaining popularity over the past few years, with its adopters claiming that it helps them write faster, memory efficient, and more reliable software.

Depending on your chosen programming language and stack, you may have never used a profiler in production. The very idea of using a profiler in production for a web service may seem unrealistic, due to the amount of overhead involved. After all, aren’t profilers extremely computationally expensive to run?

Datadog customers send trillions of data points per day. These data points are processed by Datadog and used to debug production issues in real time. But, in order to reason about all this data, we humans need visual representations. Visualizations can help us discover connections and problem points.

Datadog product designers and frontend developers have been working together to create a new, better UX for creating dashboards, which is one of the most important parts of using Datadog. A central part of this effort was building a new layout engine. Working on this project was a bit different from the usual feature work, so the colla...

As engineers, as we scale our applications and infrastructure, we accept that failure can and will happen. But, how can we get ahead of those potential failures? Gamedays are events which aim to test the resilience of a system when facing abnormal and turbulent situations, checking whether our expectations on how it will fail (or not) ...

As you scale your applications, remaining resilient to underlying network failures, resource constraints introduced by other applications, or spikes in traffic can become exponentially more complex, even with very thorough testing and processes. Chaos engineering is a discipline that encourages experimenting in production and injecting...

At Datadog, customer trust and data security are of the utmost importance.

To make sure that customers are getting the most out of the platform in the least amount of time, Datadog maintains more than 400 built-in integrations. These integrations collect metrics, events, and logs from a diverse set of sources: databases, source control, bug tracking tools, cloud providers, automation tools, and more.

eBPF (extended Berkeley Packet Filter) is a Linux technology that can run sandboxed programs in the kernel without changing kernel source code or loading kernel modules. While the kernel is an ideal place to implement monitoring/observability, networking, and security it wasn't until the recent broad adoption of eBPF that it was feasib...

The Datadog Security Platform team leverages Serverless to ingest security events across many different cloud providers, deployment platforms, and devices. These security events are then transformed and shipped to a data lake to help defend and protect the platform as a whole. Once there, these ingested events are used to drive interna...

With many blog posts published and talks given on the topic, it’s no secret that Datadog is running Kubernetes at scale. We currently run dozens of clusters, some of them with thousands of nodes. Additionally, we have clusters running in multiple clouds. How are we monitoring all of that, ensuring we can scale up quickly and safely?

Over 800 Engineers at Datadog do thousands of deployments per day, to hundreds of services in different environments, regions, and cloud providers. How can we manage all those deployments in a common way and have a reliable paper trail way to audit any changes?

Datadog is a monitoring and analytics platform that ingests trillions of data points per day, coming from more than 8,000 customers. With a complex distributed architecture and hundreds of deployments per day, needless to say sometimes things don't go as planned. Our teams have been improving the way incidents are managed at Datadog ov...

Datadog is a monitoring and analytics platform that ingests trillions of data points per day, coming from more than 8,000 customers. Each of those is associated with metadata, mostly in the form of tags, and it can also be part of streams of related data points, which can then be explored, queried, or aggregated. RocksDB is used by man...

In this session, we’ll speak with two engineers responsible for scaling the Kafka infrastructure within Datadog, Balthazar Rouberol and Jamie Alquiza. They'll share their strategy in scaling Kafka, how it’s been deployed on Kubernetes, and introduce kafka-kit; our open source toolkit for scaling Kafka clusters.

When 2 years ago Datadog decided to move its infrastructure platform to Kubernetes we didn’t expect to find so many roadblocks, but ingesting trillions of datapoints per day in a reliable fashion requires pushing the limits of cloud computing.