October 2013 Summaries

Sentry is an event logging platform that automates error monitoring, recognition, and tracking by sending notification alerts to teams as soon as exceptional events are detected. This allows developers to focus their debugging efforts on the most critical issues, streamlining the entire debugging process. With Sentry's integration with Datadog, developers can search, comment, and correlate errors with metrics and data points from other systems, creating synthetic error dashboards that provide a better understanding of unusual system performance. By adding the Sentry integration to their Datadog account, teams can quickly resolve issues faster and improve their debugging experience.

At Velocity 2013 in New York City, attendees had the opportunity to network with cutting-edge developers, architects, and system administrators. The event also featured presentations from Datadog's CTO, Alexis Lê-Quôc, and Director of Technical Operations, Mike Fiedler. Additionally, Alexis participated in an interview with O’Reilly Media’s Scott Hendrickson, discussing the history behind Datadog's name, user experience, and future plans.

Alexis Lê-Quôc, Datadog's CTO, shared insights into the company's history and future at Velocity 2013 in New York City, including explaining the origin of the name "Datadog" and offering a glimpse into its user experience and future plans. The event also provided opportunities for meeting with cutting-edge developers and speaking with product teams from integration partners. Additionally, both Alexis Lê-Quôc and Mike Fiedler delivered presentations, with one focusing on reducing signal noise and another on navigating the future of technical operations.

AWS customers experienced partial network outages last month that affected an availability zone's connectivity to the internet. The failures highlighted the importance of monitoring key metrics and being prepared for infrastructure failures. Monitoring error distributions, especially outliers, can help detect "grey" partial failures. Additionally, relying on already-deployed infrastructure in another zone or region is crucial when dealing with shared infrastructure issues. Building for failure and having a contingency plan are essential to minimize the impact of such incidents.

AWS recently experienced two significant failures: a partial network outage that affected an availability zone's connectivity to the internet, and another gray failure with a severe impact on well-known internet services. These incidents highlight the importance of monitoring key metrics, such as latency and throughput, to detect potential issues before they escalate. The analysis also emphasizes the need for contingency planning, including infrastructure backup strategies and APIs that can handle increased traffic. Additionally, building resilience into infrastructure is crucial, especially when relying on shared cloud services. By learning from these failures, organizations can improve their disaster recovery plans and reduce downtime.

The use of a cloud-based service like Amazon Web Services (AWS) or HP Public Cloud can provide metrics via their virtualization hypervisor to an API or Dashboard system, but these metrics are only part of the whole picture. The information provided is usually not all-encompassing and produced at intervals that do not lend themselves to the granularity desired when debugging performance issues. Agent-based monitoring offers faster results and higher resolution by providing direct submission methods, allowing for metrics like memory consumption by process and arriving in a matter of seconds. This type of monitoring also provides visibility under the hood, enabling users to see how the engine is working and what's consuming resources, which is essential for those who want to see detailed stats on what's running on their servers. Additionally, Datadog's Agent is fully open source, allowing users to read through every line of code before installation, with various deployment methods available for popular configuration management solutions.

Datadog's new heatmaps provide an instant and actionable way to visualize server performance across multiple servers, allowing users to quickly identify patterns and outliers in their infrastructure. By adding a third dimension of color brightness, these heatmaps enable users to distinguish between normal and anomalous behavior, making it easier to troubleshoot issues and gain insights into the overall health of their cloud-based infrastructure. With this visualization tool, users can quickly get an at-a-glance understanding of how most servers are performing and identify misbehaving servers that may be causing problems.