Azure 时序见解预览版用例Azure Time Series Insights Preview use cases

本文汇总了 Azure 时序见解预览版的几个常见用例。This article summarizes several common use cases for Azure Time Series Insights Preview. 本文中的建议可以作为使用时序见解开发应用程序和解决方案的一个起点。The recommendations in this article serve as a starting point to develop your applications and solutions with Time Series Insights.

具体而言,本文将解答以下问题:Specifically, this article answers the following questions:

以下部分描述了这些使用方案的概述。An overview of these use scenarios is described in the following sections.

简介Introduction

Azure 时序见解是一种端到端的平台即服务产品/服务。Azure Time Series Insights is an end-to-end, platform-as-a-service offering. 它用于收集、处理、分析和查询高度情景化且优化了时序的 IoT 规模的数据。It's used to collect, process, store, analyze, and query highly contextualized, time-series-optimized IoT-scale data. 时序见解是即席数据浏览和运营分析的理想选择。Time Series Insights is ideal for ad-hoc data exploration and operational analysis. 时序见解是唯一可扩展且自定义的服务产品/服务,满足行业 IoT 部署的广泛需求。Time Series Insights is a uniquely extensible, customized service offering that meets the broad needs of industrial IoT deployments.

数据浏览和直观异常检测Data exploration and visual anomaly detection

即时浏览和分析数十亿事件,确定其中的异常,发现数据的隐藏趋势。Instantly explore and analyze billions of events to spot anomalies and discover hidden trends in your data. 时序见解可为 IoT 和 DevOps 分析工作负荷提供近实时的性能。Time Series Insights delivers near real-time performance for your IoT and DevOps analysis workloads.

数据资源管理器Data explorer

大多数客户都认为,获得见解所需的最短时间是时序见解的突出功能之一:Most customers agree that the minimal amount of time required to gain insight is one of the standout features of Time Series Insights:

  • 时序见解无需前期数据准备。Time Series Insights requires no upfront data preparation.
  • 时序见解可在几分钟内快速将你连接到 Azure IoT 中心或 Azure 事件中心实例中的数十亿个事件。It works fast to connect you to billions of events in your Azure IoT Hub or Azure Event Hubs instances in minutes.
  • 连接后,可直观显示和分析数十亿事件,发现异常,发现数据的隐藏趋势。Once connected, you can visualize and analyze billions of events to spot anomalies and discover hidden trends in your data.

时序见解直观且易于使用。Time Series Insights is intuitive and simple to use. 连一行代码也无需编写即可与数据交互。You can interact with your data without writing a single line of code. 尽管时序见解为熟悉 SQL 的高级用户提供了一种基于文本的细粒度查询语言,但你也不需要学习任何新语言。There’s also no new language you're required to learn, although Time Series Insights provides a granular text-based querying language for advanced users who are familiar with SQL. 为初学者提供“选择+单击”式探索模式。It also provides select-and-click exploration for novices.

客户可以利用这种高效率,快速诊断与资产相关的问题。Customers can take advantage of the speed to diagnose asset-related issues quickly. 时序见解可以执行 DevOps 分析来找出 IoT 解决方案中 bug 的根本原因。They can perform DevOps analysis to get to the root cause of a bug in an IoT solution. 作为数据科学计划的一部分,时序见解还可以识别要标记的区域,以便进一步调查。They also can identify areas to flag for further investigation as part of their data science initiatives.

与时序见解中存储的数据进行交互主要有三种方式:There are three primary ways to interact with data stored in Time Series Insights:

  • 第一个也是最简单的方法是使用时序见解预览版资源管理器。The first and easiest way to get started is with the Time Series Insights Preview explorer. 可以使用它在一个位置快速显示所有 IoT 数据。You can use it to quickly visualize all of your IoT data in one place. 它提供热度地图等工具,可帮助你发现数据中的异常情况。It provides tools like the heat map to help you spot anomalies in your data. 它还提供透视视图。It also provides a perspective view. 使用它可以在一个仪表板中比较一个或多个时序见解环境中的多达四个视图。Use it to compare up to four views from one or more Time Series Insights environments in a single dashboard. 通过仪表板,可以查看所有位置的时序数据。The dashboard gives you a view of your time-series data across all your locations. 要规划时序见解环境,请阅读时序见解规划To plan out your Time Series Insights environment, read Time Series Insights planning.

  • 第二种开始方法是使用 JavaScript SDK 在 Web 应用程序中快速嵌入功能强大的图表和图。The second way to start is to use the JavaScript SDK to quickly embed powerful charts and graphs in your web application. 只需几行代码,即可编写功能强大的查询。With just a few lines of code, you can author powerful queries. 可使用这些查询填充折线图、饼图、条形图、热度地图和数据网格等。Use them to populate line charts, pie charts, bar charts, heat maps, data grids, and more. 通过使用 SDK,所有这些元素都是现成可用的。All of these elements exist out-of-the-box by using the SDK. SDK 还提取时序见解查询 API。The SDK also abstracts Time Series Insights query APIs. 可以使用它们来创建类似 SQL 的谓词,用于查询要在仪表板上显示的数据。You can use them to author SQL-like predicates to query the data you want to show on a dashboard. 对于混合表示层解决方案,时序见解提供参数化 URL。For hybrid presentation-layer solutions, Time Series Insights offers parameterized URLs. 它们提供与时序见解预览版资源管理器之间的无缝连接点,可帮助深入分析数据。They provide seamless connection points with the Time Series Insights Preview explorer for deep dives into data.

  • 第三种方法是使用功能强大的 API 查询存储在时序见解中的数据。The third way to start is to use the powerful APIs to query data stored in Time Series Insights. 时序见解具有 fromtofirstlast 等时态运算符。Time Series Insights has temporal operators such as from, to, first, and last. 它还具有 averageminmaxsplit byorder byDateHistogram 等聚合与转换。It has aggregations and transformations such as average, min, max, split by, order by, and DateHistogram. 它还具有 hasinandorgreater thanREGEX 等筛选运算符。It also has filtering operators such as has, in, and, or, greater than, and REGEX. 所有这些运算符使下游应用程序能够快速找到数据中的相关趋势和模式。All these operators enable downstream applications to quickly find interesting trends and patterns in your data. 可使用它们填充自生成的可视化效果以发现异常。Use them to populate homegrown visualizations to spot anomalies.

操作性分析和提高处理效率Operational analysis and driving process efficiency

使用时序见解可大规模监视设备的运行状况、使用情况和性能。Use Time Series Insights to monitor the health, usage, and performance of equipment at scale. 时序见解提供了一种衡量操作效率的简便方法。Time Series Insights provides an easy way to measure operational efficiency. 时序见解有助于管理多种不可预测的 IoT 工作负荷,且不影响引入或查询性能。Time Series Insights helps manage diverse and unpredictable IoT workloads without sacrificing ingestion or query performance.

概述Overview

如果与正确的技术或解决方案相结合,来自操作过程的数据的流式传输和连续处理可成功地转换任何业务。Streaming and continuous processing of data coming from operational processes can successfully transform any business if it's combined with the right technology or solution. 这些解决方案通常是多个系统的组合。Often these solutions are a combination of multiple systems. 它们可以探索和分析不断变化的数据,特别是在 IoT 领域,并且采用相同的模式。They enable exploration and analysis of data that changes constantly, especially in the IoT realm, and share a common pattern.

这些模式通常始于支持 IoT 的平台,这类平台从各种语言环境的设备和传感器中引入数十亿个事件。These patterns often start with IoT-enabled platforms that ingest billions of events from devices and sensors that span various locales. 这些系统可以处理和分析流数据,获取实时见解和操作。These systems process and analyze streaming data to derive real-time insights and actions. 通常将数据存档到热存储和冷存储,以实现近实时和批量分析。Data is typically archived to warm and cold store for near real-time and batch analytics.

收集的数据经过一系列处理,以便对数据进行净化并使其适合特定上下文,从而使其可用于下游查询和分析方案。Data that's collected goes through a series of processing to cleanse and contextualize it for downstream querying and analytics scenarios. Azure 提供丰富的服务,可应用于资产维护和制造等 IoT 方案。Azure offers rich services that can be applied to IoT scenarios such as asset maintenance and manufacturing. 这些服务包括时序见解、IoT 中心、事件中心、Azure 流分析、Azure Functions、Azure 逻辑应用、Azure Databricks、Azure 机器学习和 Power BI。These services include Time Series Insights, IoT Hub, Event Hubs, Azure Stream Analytics, Azure Functions, Azure Logic Apps, Azure Databricks, Azure Machine Learning, and Power BI.

解决方案体系结构可以通过以下方式实现:Solution architecture can be achieved in the following manner:

  • 通过 IoT 中心或事件中心来引入数据,以获得最佳的安全性、吞吐量和延迟。Ingest data via IoT Hub or Event Hubs for best-in-class security, throughput, and latency.
  • 执行数据处理和计算。Perform data processing and computations. 通过流分析、逻辑应用和 Azure Functions 等服务传送引入的数据。Funnel ingested data through services such as Stream Analytics, Logic Apps, and Azure Functions. 使用的服务取决于特定的数据处理需求。The service you use depends on the specific data-processing needs.
  • 来自处理管道的计算信号被推送到时序见解进行存储和分析。Computed signals from the processing pipeline are pushed to Time Series Insights for storing and analytics.

时序见解提供近实时数据浏览和基于资产的历史数据见解。Time Series Insights offers near real-time data exploration and asset-based insights over historical data. 根据业务需求,可通过将时序见解连接到 Azure HDInsight,基于存储在时序见解中的数据来运行 MapReduce 和 Hive 作业。Depending on your business needs, MapReduce and Hive jobs can run on data stored in Time Series Insights by connecting Time Series Insights to Azure HDInsight. 可通过时序见解公共图面查询 API 向 Power BI 和其他客户应用程序提供存储在时序见解中的数据。Data stored in Time Series Insights is available to Power BI and other customer applications via the Time Series Insights public surface query APIs. 此数据可用于深层业务和操作智能方案。This data can be used for deep business and operational intelligence scenarios.

高级分析Advanced analytics

与机器学习和 Azure Databricks 等高级分析服务集成。Integrate with advanced analytics services such as Machine Learning and Azure Databricks. 时序见解从数百万台设备中引入原始数据。Time Series Insights ingresses raw data from millions of devices. 它添加了可由 Azure 分析服务套件无缝使用的上下文数据。It adds contextual data that can be consumed seamlessly by a suite of Azure analytics services.

分析Analytics

高级分析和机器学习会使用和处理大量数据。Advanced analytics and machine learning consume and process large volumes of data. 该数据用于制定数据驱动的决策并执行预测分析。This data is used to make data-driven decisions and perform predictive analysis. 在 IoT 用例中,高级分析算法可以从数百万台设备中收集数据。In IoT use cases, advanced analytics algorithms learn from the data collected from millions of devices. 这些设备每秒内会多次传输数据。These devices transmit data multiple times every second. 从 IoT 设备收集的数据是原始数据。The data collected from IoT devices is raw. 它缺少上下文信息,例如设备的位置和传感器读数的单位。It lacks contextual information such as the location of the device and the unit of the sensor reading. 因此,原始数据很难直接用于高级分析。As a result, raw data is difficult to consume directly for advanced analytics.

时序见解以两种简单且经济高效的方式在 IoT 数据与高级分析之间搭建了桥梁:Time Series Insights bridges the gap between IoT data and advanced analytics in two simple and cost-effective ways:

  • 首先,时序见解通过使用 IoT 中心从数百万台设备收集原始遥测数据。First, Time Series Insights collects raw telemetry data from millions of devices by using IoT Hub. 它通过上下文信息来丰富数据,并将数据转换为 parquet 格式。It enriches data with contextual information and transforms data into a parquet format. 此格式可以轻松与其他高级分析服务集成,例如机器学习、Azure Databricks 和第三方应用程序。This format can easily integrate with other advanced analytics services, such as Machine Learning, Azure Databricks, and third-party applications.

    时序见解可以作为整个组织中所有数据的真实来源。Time Series Insights can serve as the source of truth for all data across an organization. 它创建了一个供下游分析工作负荷使用的中央存储库。It creates a central repository for downstream analytics workloads to consume. 由于时序见解是一种近实时存储服务,因此高级分析模型可以利用传入的 IoT 遥测数据不断学习。Because Time Series Insights is a near real-time storage service, advanced analytics models can learn continuously from incoming IoT telemetry data. 这样一来,模型可以更准确地进行预测。As a result, the models can make more accurate predictions.

  • 其次,机器学习和预测模型的输出可以输入到时序见解中,以直观显示并存储其结果。Second, the output of machine learning and prediction models can be fed into Time Series Insights to visualize and store their results. 此过程可帮助组织优化和调整其模型。This procedure helps organizations to optimize and tweak their models. 使用时序见解,可轻松直观显示与训练模型输出相同平面上的流式遥测数据。Time Series Insights makes it simple to visualize streaming telemetry data on the same plane as the trained model outputs. 通过这种方式,它可以帮助数据科学团队发现异常并识别模式。In this way, it helps data science teams spot anomalies and identify patterns.

后续步骤Next steps