October 2021 Summaries

Distributed traces are crucial for observing distributed systems and microservices, but storage limitations necessitate sampling methods like head-based and tail-based sampling, which have their drawbacks. Head-based sampling involves random decisions at the root span, leading to potential under-sampling of rare traces, while tail-based sampling offers more intelligent decisions at the cost of increased memory and network overhead. To address these limitations, Dynatrace is exploring partial trace sampling, which adjusts sampling rates based on the frequency of trace parts, ensuring that less frequent trace sections are sampled more often, providing a more balanced view. Although this approach results in partially sampled traces, the data is still valuable for specific queries. The concept is part of ongoing research and is influencing the development of a new OpenTelemetry sampling specification.

Dynatrace Managed release version 1.230 includes significant updates and enhancements, focusing on improving system performance and resolving vulnerabilities. Key improvements include reducing Elasticsearch backup snapshots retention to five days, updating hardware requirements for managed environments, and enhancing the Problems API to include event evidence data. Notable changes involve the exclusion of svchost.exe from injection to prevent Windows Server 2019 boot timeout issues and the introduction of management zone-aware service-level objectives. The update also allows for cloning service-level objectives and managing outage-handling settings via the Settings API. Additionally, real user traffic is now visible on the browser monitor details page, Prometheus metrics consumption behavior has been adjusted, and a new self-monitoring environment has been established for aggregating self-monitoring metrics, which is excluded from usage consumption. Several vulnerabilities, including those related to the Log4j library, have been addressed, with updates to the JVM parameters for Dynatrace Server and Elasticsearch and the removal of specific classes from the Log4j library. The release also resolves numerous issues across various components, such as cluster management, infrastructure monitoring, and user session displays, while enhancing the performance and accuracy of existing features.

Dynatrace uses its deterministic AI, Davis, to monitor Kubernetes workloads and recently identified a problem in a Keptn instance due to a 33% increase in failure rate in the mongodb-datastore workload. The AI not only alerts teams to anomalies by automatically baselining service endpoints but also identifies root causes by analyzing container logs, as demonstrated by detecting an unhandled error in the code interacting with a MongoDB instance. This capability allows for rapid problem resolution, as evidenced by the Keptn development team's swift fix after receiving detailed insights, including log data and distributed traces, which provide additional context like timing and API endpoints. The blog post underscores the effectiveness of Dynatrace's monitoring tools in enhancing operational efficiency and problem-solving speed.

The release notes for Dynatrace OneAgent version 1.227, published on October 18, 2021, outline various updates, compatibility changes, and resolved issues across multiple components and technologies. This release introduces enhanced support for technologies such as Red Hat's ABRT, Kafka, Spring Cloud Stream Kafka Binder, and OpenTelemetry, alongside the availability of oneagentzos-R12270.pax for z/OS systems. It marks the last version to support specific OpenTelemetry versions for Go and announces future deprecations of support for certain operating systems, including Amazon Linux 2, Debian 10, and Windows Server 2012 R2, effective from 2026. Improvements are made to the zDC module for faster startup on z/OS, while multiple vulnerabilities and bugs are resolved across Java, Node.js, .NET, and JavaScript modules, enhancing the stability and security of the OneAgent platform. The update also includes better Kubernetes detection, log processing enhancements, and refined handling of mobile and JavaScript functionalities, ensuring a smoother user experience across diverse environments.

Function-as-a-service (FaaS) platforms, like Google Cloud Functions (GCF), provide developers with a serverless environment to run code efficiently and with low overhead, enhancing the agility of delivering value to customers. GCF, a specialized service within the Google Cloud Platform (GCP), allows for the creation of microservices that can be triggered by events and integrate with various cloud services, supporting languages such as Node, Python, and Java. While GCF facilitates scalability and cost-effectiveness by charging only for usage, it can pose observability challenges in multicloud environments due to its nature of spinning up independent instances. Observability tools like Dynatrace offer enhanced solutions by providing end-to-end monitoring and distributed tracing, which are essential for understanding dependencies and ensuring optimal performance across a multicloud stack. Despite its benefits, GCF might not be ideal for memory-intensive or infrequent tasks due to potential delays and cost considerations, making it crucial for teams to consider alternatives like VM-based platforms in such cases.

The text discusses the transition many companies are making towards serverless cloud applications, highlighting Amazon Web Services (AWS) as a key provider of serverless solutions. While serverless architectures offer benefits such as simplicity, reliability, speed, and scalability by offloading server management to providers, they also pose challenges in observability due to the complexity of interconnected services. AWS offers a variety of serverless solutions across compute, application integration, and data storage, including services like AWS Lambda, Fargate, and Amazon S3. These services enable businesses to build efficient, event-driven applications and streamline data processing. However, maintaining observability in such environments can be challenging, which is where Dynatrace’s Software Intelligence Platform comes in, providing enhanced visibility and AI-driven analysis to simplify and optimize cloud environments. The text concludes by inviting readers to learn more about achieving comprehensive observability in AWS serverless initiatives through a webinar.

Modern applications increasingly rely on third-party services to provide seamless user experiences, making API monitoring vital for DevOps teams to ensure these services function correctly. An API, or application programming interface, facilitates communication between different software components, and API monitoring involves collecting data on API performance to identify issues affecting users. This practice is essential in multicloud environments where numerous microservices can impact an application's overall performance. Monitoring APIs enables early detection of problems, performance optimization, security compliance, and effective dependency management, while providing business insights and ensuring adherence to service-level agreements (SLAs). There are two main API monitoring methods: synthetic monitoring, which simulates user behavior to assess performance, and real user monitoring (RUM), which evaluates actual user experiences. Both methods provide comprehensive insights into API performance, allowing developers to prevent and address issues efficiently. Selecting an API monitoring tool requires considering its ability to analyze data comprehensively, support different monitoring methods, and identify third-party API impacts. Effective API monitoring, complemented by testing, ensures the development of robust applications that meet business objectives and key performance indicators (KPIs).

In response to the rapid expansion of digital services, Dynatrace and ServiceNow have formed a strategic partnership to address key challenges faced by IT teams, such as IT complexity, lack of visibility, and rising user expectations. This collaboration aims to simplify IT complexity through automatic tracing and code-level insights, provide complete visibility with deterministic AI for precise root-cause analysis, and enhance user experience with real-user insights for proactive improvements. The integration between Dynatrace and ServiceNow helps organizations reduce enterprise silos, oversee the service infrastructure in real-time, and manage change towards automation, thus enhancing decision-making and accelerating business outcomes. Key integration methods include incident integration, CMDB integration, and events integration, which collectively ensure better observability, seamless digital transformation, and uninterrupted services, ultimately driving higher-value initiatives and customer satisfaction.