March 2017 Summaries

SLAs (Service Level Agreements) are concrete representations of a company's goal to improve the performance and reliability of its services. Defining and maintaining clear objectives benefits everyone involved, from service builders to users. To collect data for SLA definition, one should gather performance metrics from user-facing applications and resources/subcomponents within internal and external applications. Synthetic monitoring and APM (Application Performance Monitoring) provide different perspectives on application performance, helping establish a clear idea of expected performance levels. Analyzing subcomponent metrics is essential for defining internal SLOs, which serve as expectations between teams within the same organization. Customizable dashboards with service-level indicators can help gauge performance and identify potential issues. Implementing SLO-driven alerts and tracking their status in real-time allows businesses to quickly address SLA breaches. With a monitoring platform like Datadog APM, one can investigate issues across various services, determine if problems are transient or require changes, and put the SLA strategy into action by setting up dashboards and alerts that reflect key resources and services.

Jon Rovello emphasizes the importance of fostering a company culture that prioritizes employee comfort and input rather than enforcing mandatory events that may not cater to everyone's interests or schedules. At Datadog's Boston office, he implemented a "Culture Club" composed of representatives from each team to ensure that events and initiatives are planned collaboratively, reflecting the diverse needs and preferences of the workforce. This approach led to the organization of engaging, optional events, such as a memorable party at the New England Aquarium, which saw high participation due to its alignment with the employees' interests. By focusing on creating an inclusive and voluntary environment, the company aims to cultivate a welcoming culture that enhances the overall employee experience, demonstrating that comfort is key to successful team bonding and engagement.

Apache Mesos is a data center abstraction layer that simplifies running distributed systems at scale, but it can be complex to deploy and requires finding compatible solutions for various tasks. To simplify the process, DC/OS (Datacenter Operating System) is introduced as a full-fledged operating system for Mesos clusters, bundled with technologies to handle service discovery, load balancing, monitoring, and more. Datadog, a popular choice for monitoring Mesos clusters, can be easily deployed across Mesos clusters using DC/OS, and out-of-the-box integrations are available for Mesos, Docker, and related services. The installation process involves installing the Datadog package on agent nodes first, followed by leader nodes, where the Datadog Agent is installed as a standalone Docker container. Once installed, users can monitor all cluster metrics, including those from Mesos, Docker, and other services, using pre-built dashboards and custom integrations. Additionally, Autodiscovery enables automatic detection of running services on each node, simplifying configuration and setup.

The post, part of a series on monitoring Apache performance, focuses on collecting key Apache metrics using the status module (mod_status) and the server access log. It provides detailed instructions on enabling and configuring mod_status, emphasizing the importance of saving configuration backups before making changes. The post also explains how to enable ExtendedStatus for additional metrics, though it may slightly impact performance under heavy load. Furthermore, it covers customizing access logs with variables for more relevant data and introduces open-source tools like Apachetop and Atop for real-time command-line monitoring. While these tools provide immediate insights, the post suggests using more comprehensive systems like Datadog for visualizing metrics, analyzing trends, and setting alerts, with a promise of integration details in the next installment of the series.

The blog post, part of a three-part series on monitoring Apache performance, explores how the Apache HTTP Server functions and identifies key performance metrics critical for efficient operation. Apache, an open-source web server, is praised for its customizability and modular architecture, which facilitates diverse use cases and contributions without disrupting core functionalities. However, its one-process-per-connection model can struggle with high concurrency, prompting interest in other web servers like NGINX and IIS. Important metrics for monitoring include request processing time, rate of requests, resource utilization, and error rates, all retrievable via Apache's mod_status and access logs. The post also discusses the Multi-Processing Modules (MPMs) in Apache—prefork, worker, and event—which dictate resource management and influence performance metrics, particularly under different operating systems. The series aims to guide users in collecting and analyzing these metrics to preempt and address performance bottlenecks, with subsequent parts detailing data collection and monitoring techniques.

The text is the final installment of a three-part series focusing on monitoring Apache performance using Datadog. It discusses setting up Datadog’s Apache integration to automatically collect and analyze Apache metrics and logs, which allows users to monitor performance both in real-time and historically. The article outlines steps to configure Apache and install the Datadog Agent, including configuring custom log formats for enhanced data collection. Additionally, it highlights how Datadog’s comprehensive integration capabilities can contextualize Apache metrics within the larger infrastructure, using dashboards and alerts to manage performance issues effectively. The series emphasizes the potential to improve monitoring by leveraging Datadog’s extensive integrations and analytical tools to gain deeper insights into server health and performance.

Datadog introduces new composite monitors that allow users to combine two or more separate monitors using logical operators to create more refined alerts. Composite monitors help reduce noise by triggering only when specific conditions are met, such as a combination of performance indicators. This feature can be used to alert on any combination of metrics and events, adding critical nuance to alert definitions. Setting up composite monitors is simple and involves selecting alert conditions, notification channels, and creating an alert message. By using composite monitors, users can receive fewer, richer, and more actionable alerts for their infrastructure and applications.

Datadog has introduced new composite monitors that allow users to combine multiple separate monitors using logical operators, enabling more refined and actionable alerting without unnecessary noise. These composite monitors can be used to create custom alerts based on combinations of performance indicators or to add nuance to existing alert definitions. By setting up a composite monitor, users can disable notifications from original singular monitors, reducing the total number of alerts received. The process of creating a composite monitor is straightforward and involves selecting alert conditions, notification channels, and defining trigger conditions using logical operators. This feature is designed to provide users with fewer, richer, and more actionable alerts, making it easier to identify potential issues in their applications and infrastructure.

This article discusses the resource and performance metrics that can be obtained from Google Compute Engine (GCE), an infrastructure-as-a-service platform within the Google Cloud Platform. The majority of available metrics are related to low-level resources, such as CPU utilization and network throughput. These metrics can generally be broken down into three categories: instance metrics, disk metrics, and network metrics. Additionally, project metrics provide information on resource consumption/growth over time and anticipate potential future issues. The article also mentions that memory metrics are not available without using a third-party tool.

Google Compute Engine (GCE) is a fully managed infrastructure-as-a-service platform that enables users to spin up virtual machines on demand. It powers high-profile businesses such as Philips, Evernote, and HTC. GCE provides a range of metrics related to low-level resources, including CPU utilization, network throughput, memory utilization, and disk read/write bytes. These metrics can be categorized into three main areas: instance metrics, which shed light on resource utilization at the individual host level; network metrics, which monitor traffic and packet drops; and project metrics, which track resource consumption and growth over time. Understanding these metrics is crucial for diagnosing performance issues in hosted applications and identifying friction points in the infrastructure.

Evan Mouzakitis provides a detailed guide on collecting and monitoring metrics from Google Compute Engine (GCE) using Google's tools, as part of a three-part series. This installment focuses on the use of Google Cloud Platform's Stackdriver monitoring API to extract GCE performance metrics, requiring OAuth2.0 authentication via service or user account authorization. The post explains how to set up and use the API for querying timeseries data, utilizing parameters such as start and end times, filters, and aggregation options, and addresses potential issues like missing security headers. Additionally, the post highlights the gcloud command-line tool, which, while limited to quota-related metrics, is useful for managing Google Cloud resources. The guide also introduces Google's Stackdriver monitoring service, which facilitates visualization and alerting on GCE metrics through customizable dashboards and built-in alerting systems. For comprehensive monitoring and analysis of GCE and other cloud environments, the post suggests using Datadog, which offers integrations with over 850 technologies, including GCE.

Evan Mouzakitis concludes a three-part series on monitoring Google Compute Engine (GCE) by detailing how Datadog can enhance the monitoring capabilities beyond Google's native tools. While the previous parts discussed key metrics and their collection via Stackdriver Monitoring API and gcloud, this final part highlights Datadog's ability to provide comprehensive visualization, alerting, and correlation of GCE metrics with other infrastructure components. Datadog retains data at full granularity for 15 months, facilitating long-term trend analysis, and can integrate with over 850 technologies and services through its Agent. Users can quickly start monitoring by enabling the Google Cloud Platform integration or installing the Datadog Agent, which offers additional system resource metrics. Datadog's customizable dashboards and advanced tagging and alerting features enable detailed monitoring and proactive issue detection, providing greater observability into GCE operations.

CloudCheckr is a web-based platform that helps businesses monitor and optimize their AWS infrastructure by providing customized recommendations based on resource consumption data. By integrating with Datadog, users can access granular memory usage metrics for every AWS instance, which are used to inform CloudCheckr's analyses and generate Right-Sizing Reports. These reports provide best practice recommendations for rebalancing and optimizing instances while saving costs. Users can also set custom alerts in CloudCheckr to track consumption or spending. To get started, simply add Datadog API and application keys to the CloudCheckr Extensions.

CloudCheckr is a web-based platform that allows users to monitor and optimize the cost and performance of their AWS infrastructure, providing customized recommendations. By integrating with Datadog, CloudCheckr offers granular memory usage metrics for every AWS instance, enabling data-driven decisions on resource management and cost optimization. The platform provides historical data, performance trends, and right-sizing reports to help users maintain an agile and cost-effective infrastructure. Users can set custom alerts to track their consumption or spending, and integrate Datadog with CloudCheckr through a simple process that requires only API keys. With its integration, users can quickly make informed decisions on resource allocation and optimize their AWS usage.