如何:管理数据馈送How to: Manage your data feeds

了解如何在指标顾问中管理已载入的数据馈送。Learn how to manage your onboarded data feeds in Metrics Advisor. 本文将指导你完成在指标顾问中管理数据馈送的步骤。This article guides you through managing data feeds in Metrics Advisor.

编辑数据馈送Edit a data feed


创建数据馈送后,无法更改以下详细信息。The following details cannot be changed after a data feed has been created.

  • 数据馈送 IDData feed ID
  • 创建时间Created Time
  • 维度Dimension
  • 源类型Source Type
  • 粒度Granularity

仅允许数据馈送的管理员对其进行更改。Only the administrator of a data feed is allowed to make changes to it.

若要暂停或重新激活数据馈送:To pause or reactivate a data feed:

  1. 在“数据馈送列表”页上,单击要对数据馈送执行的操作。On the data feed list page, click the operation you want to perform on the data feed.

  2. 在“数据馈送详细信息”页上,单击“状态”切换按钮。On the data feed details page, click the Status switch button.

若要删除数据馈送:To delete a data feed:

  1. 在“数据馈送列表”页上,单击数据馈送上的“删除”。On the data feed list page, click Delete on the data feed.

  2. 在“数据馈送详细信息”页上,单击“删除”。In the data feed details page, click Delete.

更改开始时间时,需要再次验证架构。When changing the start time, you need to verify the schema again. 可以使用“编辑参数”对其进行更改。You can change it by using Edit parameters.

回填数据馈送Backfill your data feed

选择“回填”按钮触发时间戳的即时引入,以修复失败的引入或替代现有数据。Select the Backfill button to trigger an immediate ingestion on a time-stamp, to fix a failed ingestion or override the existing data.

  • 开始时间为包含时间。The start time is inclusive.
  • 结束时间为非包含时间。The end time is exclusive.
  • 异常检测仅在所选范围内重新触发。Anomaly detection is re-triggered on selected range only.


管理数据馈送的权限Manage permission of a data feed

工作区访问权限由指标顾问资源控制,该资源使用 Azure Active Directory 进行身份验证。Workspace access is controlled by the Metrics Advisor resource, which uses Azure Active Directory for authentication. 另一层权限控制应用于指标数据。Another layer of permission control is applied to metric data.

利用指标顾问可以向不同的人员组授予针对不同数据馈送的权限。Metrics Advisor lets you grant permissions to different groups of people on different data feeds. 有两种类型的角色:There are two types of roles:

  • 管理员:拥有管理数据馈送的完全权限,包括修改和删除。Administrator: Who has full permissions to manage a data feed, including modify and delete.
  • 查看者:有权访问数据馈送的只读视图。Viewer: Who has access to a read-only view of the data feed.

高级设置Advanced settings

创建新的数据馈送时有几个可选的高级设置,可以在数据馈送详细信息页中修改这些设置。There are several optional advanced settings when creating a new data feed, they can be modified in data feed detail page.

引入选项Ingestion options

  • 引入时间偏移:默认情况下,根据指定的粒度对数据进行引入。Ingestion time offset: By default, data is ingested according to the specified granularity. 例如,时间戳为每日的指标将在其时间戳的后一天引入。For example, a metric with a daily timestamp will be ingested one day after its timestamp. 可以通过偏移量使用正数延迟引入时间,或使用负数提前引入时间 。You can use the offset to delay the time of ingestion with a positive number, or advance it with a negative number.

  • 最大并发:如果数据源支持有限的并发,请设置此参数。Max concurrency: Set this parameter if your data source supports limited concurrency. 否则保留默认设置。Otherwise leave at the default setting.

  • 多长时间后停止重试:如果数据引入失败,它将在一段时间内自动重试。Stop retry after: If data ingestion has failed, it will retry automatically within a period. 该时段的开始是发生第一次数据引入的时间。The beginning of the period is the time when the first data ingestion happened. 该时段的长度根据粒度定义。The length of the period is defined according to the granularity. 如果保留默认值 (-1),则将根据粒度确定该值,如下所示。If leaving the default value (-1), the value will be determined according to the granularity as below.

    粒度Granularity 多长时间后停止重试Stop Retry After
    每日、自定义(>= 1 天)、每周、每月、每年Daily, Custom (>= 1 Day), Weekly, Monthly, Yearly 7 天7 days
    每小时、自定义(< 1 天)Hourly, Custom (< 1 Day) 72 小时72 hours
  • 最小重试间隔:重试从源拉取数据时,可以指定最小间隔。Min retry interval: You can specify the minimum interval when retrying pulling data from source. 如果保留默认值 (-1),则将根据粒度确定重试间隔,如下所示。If leaving the default value (-1), the retry interval will be determined according to the granularity as below.

    粒度Granularity 最小重试间隔Minimum Retry Interval
    每日、自定义(>= 1 天)、每周、每月Daily, Custom (>= 1 Day), Weekly, Monthly 30 分钟30 minutes
    每小时、自定义(< 1 天)Hourly, Custom (< 1 Day) 10 分钟10 minutes
    每年Yearly 1 天1 day

检测时填充间隙:Fill gap when detecting:


此设置不会影响数据源,也不会影响门户上显示的数据图表。This setting won't affect your data source and will not affect the data charts displayed on the portal. 仅在异常检测期间进行自动填充。The auto-filling only occurs during anomaly detection.

某些时序不是连续的。Some time series are not continuous. 当缺少数据点时,指标顾问将使用指定的值在异常检测之前填充它们以提高准确性。When there are missing data points, Metrics Advisor will use the specified value to fill them before anomaly detection for better accuracy. 选项包括:The options are:

  • 使用以前的实际数据点的值。Using the value from the previous actual data point. 这是默认设置。This is used by default.
  • 使用特定值。Using a specific value.

操作链接模板用于预定义可操作的 HTTP URL,其中包含占位符 %datafeed%metric%timestamp%detect_config%tagsetAction link templates are used to predefine actionable HTTP urls, which consist of the placeholders %datafeed, %metric, %timestamp, %detect_config, and %tagset. 可以使用模板从异常或事件重定向到特定 URL 以向下钻取。You can use the template to redirect from an anomaly or an incident to a specific URL to drill down.


填充操作链接后,单击事件列表的操作选项上的“转到操作链接”,然后单击事件树的右键单击菜单。Once you've filled in the action link, click Go to action link on the incident list's action option, and incident tree's right-click menu. 将操作链接模板中的占位符替换为异常或事件的相应值。Replace the placeholders in the action link template with the corresponding values of the anomaly or incident.

占位符Placeholder 示例Examples 评论Comment
%datafeed - 数据馈送 IDData feed ID
%metric - 指标 IDMetric ID
%detect_config - 检测配置 IDDetect config ID
%timestamp - 持久事件的异常或结束时间的时间戳Timestamp of an anomaly or end time of a persistent incident
%tagset %tagset,%tagset,
[ %tagset.get("Dim1", "filterVal")]
事件的异常或顶级异常的维度值。Dimension values of an anomaly or top anomaly of an incident.
filterVal 用于筛选掉方括号内的匹配值。The filterVal is used to filter out matching values within the square brackets.

示例:Examples :

  • 如果操作链接模板为 https://action-link/metric/%metric?detectConfigId=%detect_configIf the action link template is https://action-link/metric/%metric?detectConfigId=%detect_config,

    • 操作链接 https://action-link/metric/1234?detectConfigId=2345 将转到指标 1234 下的异常或事件,并检测配置 2345The action link https://action-link/metric/1234?detectConfigId=2345 would go to anomalies or incidents under metric 1234 and detect config 2345.
  • 如果操作链接模板为 https://action-link?[Dim1=%tagset.get('Dim1','')&][Dim2=%tagset.get('Dim2','')]If the action link template is https://action-link?[Dim1=%tagset.get('Dim1','')&][Dim2=%tagset.get('Dim2','')],

    • 当异常为 { "Dim1": "Val1", "Dim2": "Val2" } 时,操作链接将为 https://action-link?Dim1=Val1&Dim2=Val2The action link would be https://action-link?Dim1=Val1&Dim2=Val2 when the anomaly is { "Dim1": "Val1", "Dim2": "Val2" }.
    • 当异常为 { "Dim1": "", "Dim2": "Val2" } 时,操作链接将为 https://action-link?Dim2=Val2,因为对于维度值为空的字符串将跳过 [Dim1=***&]The action link would be https://action-link?Dim2=Val2 when the anomaly is { "Dim1": "", "Dim2": "Val2" } , since [Dim1=***&] is skipped for the dimension value empty string.
  • 如果操作链接模板为 https://action-link?filter=[Name/Dim1 eq '%tagset.get('Dim1','')' and ][Name/Dim2 eq '%tagset.get('Dim2','')']If the action link template is https://action-link?filter=[Name/Dim1 eq '%tagset.get('Dim1','')' and ][Name/Dim2 eq '%tagset.get('Dim2','')'],

    • 当异常为 { "Dim1": "Val1", "Dim2": "Val2" } 时,操作链接将为 https://action-link?filter=Name/Dim1 eq 'Val1' and Name/Dim2 eq 'Val2'The action link would be https://action-link?filter=Name/Dim1 eq 'Val1' and Name/Dim2 eq 'Val2' when the anomaly is { "Dim1": "Val1", "Dim2": "Val2" },
    • 当异常为 { "Dim1": "", "Dim2": "Val2" } 时,操作链接将为 https://action-link?filter=Name/Dim2 eq 'Val2',因为对于维度值为空的字符串将跳过 [Name/Dim1 eq '***' and ]The action link would be https://action-link?filter=Name/Dim2 eq 'Val2' when anomaly is { "Dim1": "", "Dim2": "Val2" } since [Name/Dim1 eq '***' and ] is skipped for the dimension value empty string.

“数据馈送不可用”警报设置"Data feed not available" alert settings

如果从数据馈送开始引入之时起,在指定的宽限期内没有从源中引入任何数据,则数据馈送会被视为不可用。A data feed is considered as not available if no data is ingested from the source within the grace period specified from the time the data feed starts ingestion. 这种情况会触发警报。An alert is triggered in this case.

若要配置警报,需要先创建挂钩To configure an alert, you need to create a hook first. 警报将通过配置的挂钩发送。Alerts will be sent through the hook configured.

  • 宽限期:“宽限期”设置用于确定在没有数据点引入时发送警报的时间。Grace period: The Grace period setting is used to determine when to send an alert if no data points are ingested. 参考点是第一次引入的时间。The reference point is the time of first ingestion. 如果引入失败,指标顾问将继续按照粒度指定的固定间隔进行尝试。If an ingestion fails, Metrics Advisor will keep trying at a regular interval specified by the granularity. 如果在宽限期后仍失败,则会发送警报。If it continues to fail past the grace period, an alert will be sent.

  • 自动推迟:当此选项设置为零时,每个不可用的时间戳都将触发警报。Auto snooze: When this option is set to zero, each timestamp with Not Available triggers an alert. 如果指定的设置不是零,则根据指定的设置,第一个不可用的时间戳之后的连续时间戳不会触发警报。When a setting other than zero is specified, continuous timestamps after the first timestamp with not available are not triggered according to the the setting specified.

后续步骤Next steps