analytics-v1.json

{
  "type": "kafka",
  "dataSchema": {
    "dataSource": "analytics-v1",
    "parser": {
      "type": "string",
      "parseSpec": {
        "format": "json",
        "timestampSpec": {
          "column": "timestamp",
          "format": "auto"
        },
        "dimensionsSpec": {
          "dimensions": [
            "userId",
            "eventType"
          ]
        }
      }
    },
    "metricsSpec": [
      {
        "type": "count",
        "name": "count"
      },
      {
        "type": "hyperUnique",
        "name": "userId.hll",
        "fieldName": "userId"
      }
    ],
    "granularitySpec": {
      "type": "uniform",
      "segmentGranularity": "DAY",
      "queryGranularity": "HOUR",
      "intervals": [
        "2022-03-07/2031-12-01"
      ]
    }
  },
  "tuningConfig": {
    "type": "kafka",
    "maxRowsInMemory": 50000,
    "intermediatePersistPeriod": "PT1S",
    "maxRowsPerSegment": 500000
  },
  "ioConfig": {
    "topic": "analytics-v1",
    "consumerProperties": {
      "bootstrap.servers": "kafka:9092"
    },
    "useEarliestOffset": true,
    "taskCount": 1,
    "replicas": 1,
    "taskDuration": "PT1H"
  }
}