flink 基站与服务器长连接，每次连接和断开都会上报数据，统计过去一小时每个基站断开次数和时长

宝哥大数据2025-03-31 8:58

模拟生成数据

sql 复制代码

CREATE TABLE ods_station_log (
  base_station_id int,   -- 基站ID
  event_type      int,   -- 事件类型: connect/disconnect
  event_time      TIMESTAMP_LTZ(3), -- 事件时间
  WATERMARK FOR event_time AS event_time - INTERVAL '5' SECOND -- 允许5秒乱序
) WITH (
  'connector' = 'datagen',
  
  -- 基站ID生成规则（BS001 ~ BS100）
  'fields.base_station_id.kind' = 'random',
  'fields.base_station_id.min' = '1',
  'fields.base_station_id.max' = '2',
  
  -- 事件类型随机生成（50%概率为 connect/disconnect）
  'fields.event_type.kind' = 'random',
  'fields.event_type.min' = '0',
  'fields.event_type.max' = '1',
  
  -- 事件时间生成规则（模拟1小时数据，每秒10条）
  'fields.event_time.kind' = 'random',
  'fields.event_time.max-past' = '1000s',
  -- 控制数据生成速率
  'rows-per-second' = '1'
);

需求：基站与服务器长连接，每次连接和断开都会上报数据，统计过去一小时每个基站断开次数和时长

思路：

sql 复制代码

CREATE TEMPORARY VIEW disconnect_records2 AS
SELECT
  base_station_id,
  connect_time,
  disconnect_time,
  TIMESTAMPDIFF(SECOND, connect_time, disconnect_time) AS duration
FROM ods_station_log
MATCH_RECOGNIZE (
  PARTITION BY base_station_id
  ORDER BY event_time
  MEASURES
    e_connect.event_time AS connect_time,
    e_disconnect.event_time AS disconnect_time
  ONE ROW PER MATCH
  AFTER MATCH SKIP PAST LAST ROW
  PATTERN (e_connect e_disconnect) 
  DEFINE
    e_connect AS e_connect.event_type = 1,
    e_disconnect AS e_disconnect.event_type = 0
);