包阅导读总结
1.
关键词:OpenTelemetry、GKE、OTLP 管道、收集数据、配置优化
2.
总结:文本介绍了针对 GKE 的新的精选 OpenTelemetry 摄入管道,包括其定义、组成、目标和提供的优化配置,还说明了部署和使用该管道所需的 Google Cloud 服务账号设置。
3.
主要内容:
– 介绍 OpenTelemetry 是一种用于日志、指标和跟踪的开源、无供应商依赖的可观测性标准,因资源丰富和社区支持在应用监测中发展迅速。
– 更多客户希望在 GKE 中采用 OpenTelemetry 标准,为此提供了相关应用的文档。
– 详细介绍新的 GKE 精选 OpenTelemetry 协议管道
– 定义:OTLP 是一种标准数据格式,应用可通过 OpenTelemetry Collector 处理和路由数据。
– 组成:包括开源 OpenTelemetry Collector 部署、考虑常见用例和问题的配置、必要的访问控制资源。
– 目标:提供开箱即用解决方案,可添加配置满足需求。
– 说明管道提供的样本优势
– 优化了常见 GKE 用例的配置,处理资源检测、数据批处理等场景。
– 配置经过谷歌支持和验证。
– 介绍管道的部署和使用
– 需要 Google Cloud 服务账号向 GCP 可观测性后端发送数据,提供了设置命令。
思维导图:
文章来源:cloud.google.com
作者:Mike Dame
发布时间:2024/7/31 0:00
语言:英文
总字数:783字
预计阅读时间:4分钟
评分:79分
标签:Kāi Fāng Jiǎn Cè,GKE,Guān Cè Xìng,Gǔ Gē Yún,Kū Bèi Nè Tè
以下为原文内容
本内容来源于用户推荐转载,旨在分享知识与观点,如有侵权请联系删除 联系邮箱 media@ilingban.com
OpenTelemetry is an open-source, vendor-agnostic observability standard for logs, metrics, and traces that’s quickly growing as a format for application monitoring, thanks to its collection of libraries, agents, compatible backends, and community support.
As more customers told us they wanted to adopt the OpenTelemetry standard for Google Kubernetes Engine (GKE), we provided documentation on instrumenting your applications with OpenTelemetry. Today, we’re excited to announce a new curated OpenTelemetry Protocol pipeline for GKE and documentation to simplify the process of collecting OTLP data from your instrumented applications and exporting it to Cloud Observability.
What is the OTLP pipeline?
The OpenTelemetry Protocol (OTLP) defines a standard, vendor-agnostic data format for application logs, metrics, and traces. Applications instrumented with OpenTelemetry to report these signals with OTLP can use an OpenTelemetry Collector to process and route that data.
On GKE, this means that your workload pods can be instrumented with OpenTelemetry and export that data to be handled by a Collector that sends logs, metrics, and traces to Google Cloud Observability.
The curated OTLP Ingestion Pipeline is a set of Kubernetes manifests, pre-configured to deploy the minimum necessary resources to install an OpenTelemetry Collector in your GKE cluster. It includes:
-
A deployment of the open source OpenTelemetry Collector
-
A Collector configuration designed with common GKE use cases and known issues in mind, including support for high-volume signals and resource detection
-
Necessary role-based access control resources to enable the Collector to identify workloads and communicate with GCP Observability backends
The goal of this pipeline is to provide an out-of-the-box solution to quickly get started with OTLP collection, with the ability to add configurations to suit your needs.
What this sample offers
Manually deploying and configuring an OTLP pipeline can be a tedious process. With the number of configuration options available, it’s not always clear which settings are optimal for your environment and backend.
Fortunately, most use cases on GKE can be optimized for Google Cloud Observability with a common config, which this pipeline provides. This config handles the following scenarios:
-
Resource detection, precisely identifying workloads on GKE that are the sources of telemetry
-
Batching telemetry data up to the maximum throughput allowed by Google Cloud Observability, reducing the amount of API calls made by the Collector
-
Memory limiting configured to prevent Collector OOM crashes
-
Transformations to prevent metric conflicts and ensure attribute naming in compliance with Google Cloud and OpenTelemetry standards
This default config is Google-supported, and is validated by additional tests to make sure it remains accurate and compatible with new releases of the OpenTelemetry Collector.
The manifests in this pipeline install the Collector with the above config settings in a Kubernetes Deployment, which we have specifically chosen as a good installation mode for most use cases due to its ability to offer high availability uptime alongside cluster-wide resource detection.
The full Collector config and Kubernetes manifests are available on GitHub and can be installed with the standard `kubectl` CLI using `kustomize`.
Deploying and using the pipeline
The OTLP pipeline requires a Google Cloud Service Account to send data to the GCP Observability backend. To set one up, use the following commands: