1. JSON Shape

1.1 includeMeta = false

[
  {
    "ts_ms": 1734870900000,
    "point_id": 10001,
    "point_key": "temp",
    "value": 25.6,
    "data_type": null
  },
  {
    "ts_ms": 1734870900000,
    "point_id": 10002,
    "point_key": "running",
    "value": true,
    "data_type": null
  }
]

1.2 includeMeta = true

[
  {
    "ts_ms": 1734870900000,
    "point_id": 10001,
    "point_key": "temp",
    "value": 25.6,
    "data_type": "float64"
  }
]

---

2. Applicable Scope

TimeseriesRows is mainly used for:

• Telemetry
• Attributes (Will merge three types of attributes)

For other event types, output may be empty array or ignored. Suggest still using EnvelopeJson for mixed event scenarios.

---

3. Best Practice

3.1 Partition Suggestion

Put time partition on topic layer, keep payload stable:

lake.{{yyyy}}.{{MM}}.{{dd}}.{{HH}}.telemetry

3.2 Benefits of Row-based

• Closer to "Fact Table" write form (One point value per row)
• Easier for platform side to do column mapping or tagging by point_key