你当前正在访问 Microsoft Azure Global Edition 技术文档网站。 如果需要访问由世纪互联运营的 Microsoft Azure 中国技术文档网站,请访问 https://docs.azure.cn

将时序见解 Gen1 迁移到 Azure 数据资源管理器

注意

2025 年 3 月之后,将不再支持时序见解 (TSI) 服务。 请考虑尽快将现有 TSI 环境迁移到备用解决方案。 有关弃用和迁移的详细信息,请访问我们的文档

概述

建议从事件中心或 IoT 中心中使用新的使用者组来设置 Azure 数据资源管理器群集并等待保留期过去,然后使用时序见解环境中的相同数据填充 Azure 数据资源管理器。 如果需要从时序见解环境中导出遥测数据,建议使用时序见解查询 API 分批下载事件,并以所需的格式进行序列化。 对于参考数据,可以使用时序见解资源管理器或参考数据 API 下载参考数据集,并将其作为另一个表上传到 Azure 数据资源管理器。 然后,可以使用 Azure 数据资源管理器中的具体化视图来联接参考数据和遥测数据。 将具体化视图与 arg_max() 聚合函数结合使用,该函数将获取每个实体的最新记录,如以下示例中所示。 有关具体化视图的详细信息,请阅读以下文档:具体化视图用例

.create materialized-view MVName on table T
{
    T
    | summarize arg_max(Column1,*) by Column2
}

将时序见解查询转换为 KQL

对于查询,建议在 Azure 数据资源管理器中使用 KQL。

事件

{
  "searchSpan": {
    "from": "2021-11-29T22:09:32.551Z",
    "to": "2021-12-06T22:09:32.551Z"
  },
  "predicate": {
    "predicateString": "([device_id] = 'device_0') AND ([has_error] != null OR [error_code] != null)"
  },
  "top": {
    "sort": [
      {
        "input": {
          "builtInProperty": "$ts"
        },
        "order": "Desc"
      }
    ],
    "count": 100
  }
}
	events
| where _timestamp >= datetime("2021-11-29T22:09:32.551Z") and _timestamp < datetime("2021-12-06T22:09:32.551Z") and deviceid == "device_0" and (not(isnull(haserror)) or not(isempty(errorcode)))
| top 100 by _timestamp desc

聚合

{
    "searchSpan": {
      "from": "2021-12-04T22:30:00Z",
      "to": "2021-12-06T22:30:00Z"
    },
    "predicate": {
      "eq": {
        "left": {
          "property": "DeviceId",
          "type": "string"
        },
        "right": "device_0"
      }
    },
    "aggregates": [
      {
        "dimension": {
          "uniqueValues": {
            "input": {
              "property": "DeviceId",
              "type": "String"
            },
            "take": 1
          }
        },
        "aggregate": {
          "dimension": {
            "dateHistogram": {
              "input": {
                "builtInProperty": "$ts"
              },
              "breaks": {
                "size": "2d"
              }
            }
          },
          "measures": [
            {
              "count": {}
            },
            {
              "sum": {
                "input": {
                  "property": "DataValue",
                  "type": "Double"
                }
              }
            },
            {
              "min": {
                "input": {
                  "property": "DataValue",
                  "type": "Double"
                }
              }
            },
            {
              "max": {
                "input": {
                  "property": "DataValue",
                  "type": "Double"
                }
              }
            }
          ]
        }
      }
    ]
  }

	let _q = events | where _timestamp >= datetime("2021-12-04T22:30:00Z") and _timestamp < datetime("2021-12-06T22:30:00Z") and deviceid == "device_0";
let _dimValues0 = _q | project deviceId | sample-distinct 1 of deviceId;
_q
| where deviceid in (_dimValues0) or isnull(deviceid)
| summarize
    _meas0 = count(),
    _meas1 = iff(isnotnull(any(datavalue)), sum(datavalue), any(datavalue)),
    _meas2 = min(datavalue),
    _meas3 = max(datavalue),
    by _dim0 = deviceid, _dim1 = bin(_timestamp, 2d)
| project
    _dim0,
    _dim1,
    _meas0,
    _meas1,
    _meas2,
    _meas3,
| sort by _dim0 nulls last, _dim1 nulls last