查詢優化在Dashdb

id timestamp  status 
167 6/28/2016 8:05:58 PM 0 
167 6/28/2016 8:06:02 PM 0 
167 6/28/2016 8:06:05 PM 0 
167 6/28/2016 8:08:06 PM 0 
167 6/28/2016 8:08:10 PM 0 
167 6/28/2016 8:08:13 PM 3 
167 6/28/2016 8:08:17 PM 3 
167 6/28/2016 8:10:41 PM 3 
167 6/28/2016 8:10:45 PM 3 
167 6/28/2016 8:10:48 PM 3 
167 6/28/2016 8:10:51 PM 0 
167 6/28/2016 8:10:55 PM 0 
167 6/28/2016 8:26:40 PM 0 
167 6/28/2016 8:26:44 PM 0 
167 6/28/2016 8:26:47 PM 3 
167 6/28/2016 8:26:50 PM 3 
167 6/28/2016 8:29:19 PM 3 
167 6/28/2016 8:29:23 PM 3 
167 6/28/2016 8:29:26 PM 0 
167 6/28/2016 8:29:29 PM 0 
167 6/28/2016 8:30:33 PM 0 
167 6/28/2016 8:30:37 PM 0 
167 6/28/2016 8:30:40 PM 3 
167 6/28/2016 8:30:44 PM 3 
167 6/28/2016 8:33:11 PM 3 
167 6/28/2016 8:33:15 PM 3 
167 6/28/2016 8:33:18 PM 0 
167 6/28/2016 8:33:21 PM 0 
167 6/28/2016 8:33:45 PM 0 
167 6/28/2016 8:33:48 PM 0 
167 6/28/2016 8:33:52 PM 0 
167 6/28/2016 8:33:55 PM 3 
167 6/28/2016 8:33:58 PM 3 
167 6/28/2016 8:34:59 PM 3 
167 6/28/2016 8:35:03 PM 3

以上數據是從我們IoT系統之一，它主要講述在一臺機器是在WRT什麼狀態timestamp與0 being Stop and 3 being running.查詢優化在Dashdb

我要計算在每小時生產的零件一小時，爲此邏輯是檢查機器狀態是否從3變爲0，然後產生一個aprt。

我寫了下面的查詢，然後給出預期的結果。查詢工作正常，如果我們爲3-5機器執行它，如果在24小時內對100臺機器執行相同的操作，則需要很長時間。目前，查詢每個設備需要14分鐘1小時的數據。

紅外線使用 - DashDB in bluemix，數據，這個數據庫是從cloudant再次bluemix服務推動。

==== 
select a.runstatus,a.runcount ,b.runstatus,b.runcount from (SELECT a.run_status runstatus, count(a.run_status) runcount 
FROM machinedata a where a.run_status <> (SELECT b.run_status 
     FROM machinedata b 
     WHERE a.machine_id = b.machine_id 
      AND a.timestamp > b.timestamp 
      ORDER BY b.timestamp DESC 
     LIMIT 1 
    ) and a.machine_id = 167 and a.run_status=0 AND a.`timestamp` BETWEEN '2016-06-28 20:00:01' 
            AND '2016-06-28 20:59:59' group by a.run_status) as a, 
     (SELECT a.run_status runstatus, count(a.run_status) runcount 
FROM machinedata a where a.run_status <> (SELECT b.run_status 
     FROM machinedata b 
     WHERE a.machine_id = b.machine_id 
      AND a.timestamp > b.timestamp 
      ORDER BY b.timestamp DESC 
     LIMIT 1 
    ) and a.machine_id = 167 and a.run_status=3 AND a.`timestamp` BETWEEN '2016-06-28 20:00:01' 
            AND '2016-06-28 20:59:59' group by a.run_status) as b

來源

2017-02-27 Lohith Kumar

這人會做的工作：

select timestamp part_produced_time, hour(timestamp) part_produced_hour 
     from (select *, first_value(status) over 
         (order by timestamp asc 
          rows between 1 preceding and current row) 
         as prev_status 
      from machinedata 
      ) 
     where status=0 and prev_status=3

來源

2017-02-27 12:20:08

HI託斯滕，語法「按訂單」 –

@LohithKumar給錯誤我附近再次驗證它和它的作品，因爲我粘貼它的罰款。 –

這裏是我爲我的測試創建的表'''CREATE TABLE MACHINEDATA（ID INT，TIMESTAMP TIMESTAMP，STATUS INT）'' –

查詢優化在Dashdb

回答

相關問題