GAリリースされたAmazon Timestreamにいろいろなクエリーを投げてみました。

サンプルデータ

Timestreamのデータベースを作成する際に、サンプルデータが入ったものを作成することができます。
今回はこちらを使ってクエリーを投げてみました。

fleet	truck_id	fuel_capacity	model	load_capacity	make	measure_value::double	measure_value::varchar	measure_name	time
Alpha	7128555298	100	359	1000	Peterbilt	7.842702716127531	-	fuel-reading	2020-10-09 14:11:54.267000000
Alpha	1641688615	100	359	1000	Peterbilt	-	36.1627° N, 86.7816° W	location	2020-10-09 14:11:52.086000000
Alpha	9355060257	100	Wrecker	500	Peterbilt	12.0	-	speed	2020-10-09 14:11:48.210000000
Alpha	7219245711	100	Wrecker	500	Peterbilt	19.78241022554396	-	fuel-reading	2020-10-09 14:11:44.294000000
Alpha	21135517	100	359	1000	Peterbilt	74.1987909949444	-	fuel-reading	2020-10-09 14:11:27.965000000
Alpha	433496933	100	Wrecker	500	Peterbilt	112.0	-	load	2020-10-09 14:11:27.788000000
Alpha	3496454495	150	W925	1000	Kenworth	-	36.1627° N, 86.7816° W	location	2020-10-09 14:11:27.742000000
Alpha	4354044937	100	359	1000	Peterbilt	69.50232420155284	-	fuel-reading	2020-10-09 14:11:25.140000000
Alpha	5644432615	150	W925	1000	Kenworth	101.12753874492175	-	fuel-reading	2020-10-09 14:11:24.987000000
Alpha	7602194211	100	359	1000	Peterbilt	138.0	-	load	2020-10-09 14:11:18.528000000

timeに時刻、measure_name測定値の名前が入り、measure_value::doubleまたはmeasure_value::varcharに計測値が入ってます。
fleetやtruck_idなどは属性値で、Dimensionと呼ぶようです。

SQL

いわゆる通常のSQLが使えます。サブクエリも使えるとのことです。

SELECT * 
FROM "IoT-Sample"."IoT" 
WHERE truck_id = '4132246625' AND measure_name = 'speed'
ORDER BY time DESC
LIMIT 10

fleet	truck_id	fuel_capacity	model	load_capacity	make	measure_value::double	measure_value::varchar	measure_name	time
Alpha	4132246625	100	359	1000	Peterbilt	67.0	-	speed	2020-10-09 14:09:21.208000000
Alpha	4132246625	100	359	1000	Peterbilt	73.0	-	speed	2020-10-09 12:24:41.422000000
Alpha	4132246625	100	359	1000	Peterbilt	16.0	-	speed	2020-10-09 12:16:49.933000000
Alpha	4132246625	100	359	1000	Peterbilt	3.0	-	speed	2020-10-09 12:00:57.192000000
Alpha	4132246625	100	359	1000	Peterbilt	35.0	-	speed	2020-10-09 11:51:33.847000000
Alpha	4132246625	100	359	1000	Peterbilt	49.0	-	speed	2020-10-09 11:48:21.102000000
Alpha	4132246625	100	359	1000	Peterbilt	50.0	-	speed	2020-10-09 11:47:10.982000000
Alpha	4132246625	100	359	1000	Peterbilt	52.0	-	speed	2020-10-09 11:01:30.840000000
Alpha	4132246625	100	359	1000	Peterbilt	54.0	-	speed	2020-10-09 09:22:08.533000000
Alpha	4132246625	100	359	1000	Peterbilt	19.0	-	speed	2020-10-09 09:03:11.096000000

Date / Time Functions

時系列データベースというので、時刻関係の関数が用意されており、bin(timestamp, X unit)関数はX unitに切り捨てしてくれます。
例えば30分ごとの平均を取得。

SELECT truck_id, avg(measure_value::double) as avg_speed, BIN(time, 30m) as binned_time
FROM "IoT-Sample"."IoT" 
WHERE truck_id = '4132246625' AND measure_name = 'speed'
GROUP BY truck_id, BIN(time, 30m)
ORDER BY binned_time DESC
LIMIT 10

truck_id	avg_speed	binned_time
4132246625	67.0	2020-10-09 14:00:00.000000000
4132246625	30.666666666666668	2020-10-09 12:00:00.000000000
4132246625	44.666666666666664	2020-10-09 11:30:00.000000000
4132246625	52.0	2020-10-09 11:00:00.000000000
4132246625	36.5	2020-10-09 09:00:00.000000000

Timeseries views

Timestreamでは、timeseriesというデータタイプが用意されいて、CREATE_TIME_SERIES関数を使うことで、timeseries型に変換できます。
timeseries型のデータは、時刻と計測値のペアがJSON形式で格納されます。

SELECT truck_id, CREATE_TIME_SERIES(time, measure_value::double) as speed
FROM "IoT-Sample"."IoT" 
WHERE measure_name = 'speed'
GROUP BY truck_id
ORDER BY truck_id
LIMIT 10

truck_id	speed
1234546252	[ { time: 2020-10-09 09:45:32.192000000, value: 65.0 }, { time: 2020-10-09 11:16:40.034000000, value: 45.0 }... View 8 more row(s) ]
1575739296	[ { time: 2020-10-09 09:15:07.856000000, value: 56.0 }, { time: 2020-10-09 10:29:40.381000000, value: 30.0 }... View 8 more row(s) ]
1588092325	[ { time: 2020-10-09 09:01:49.081000000, value: 19.0 }, { time: 2020-10-09 09:22:57.346000000, value: 67.0 }... View 8 more row(s) ]
1641688615	[ { time: 2020-10-09 09:49:20.010000000, value: 37.0 }, { time: 2020-10-09 10:14:42.971000000, value: 42.0 }... View 8 more row(s) ]
1682738967	[ { time: 2020-10-09 09:07:58.814000000, value: 75.0 }, { time: 2020-10-09 09:28:31.453000000, value: 8.0 }... View 8 more row(s) ]
1712492054	[ { time: 2020-10-09 09:36:27.020000000, value: 44.0 }, { time: 2020-10-09 09:49:04.039000000, value: 65.0 }... View 8 more row(s) ]
1836816173	[ { time: 2020-10-09 08:59:28.330000000, value: 29.0 }, { time: 2020-10-09 09:21:43.510000000, value: 15.0 }... View 8 more row(s) ]
199744055	[ { time: 2020-10-09 09:16:35.398000000, value: 42.0 }, { time: 2020-10-09 09:23:48.835000000, value: 46.0 }... View 8 more row(s) ]
2062792987	[ { time: 2020-10-09 09:30:31.074000000, value: 45.0 }, { time: 2020-10-09 09:54:24.573000000, value: 50.0 }... View 8 more row(s) ]
21135517	[ { time: 2020-10-09 09:23:58.552000000, value: 64.0 }, { time: 2020-10-09 10:43:06.367000000, value: 48.0 }... View 8 more row(s) ]

View X more row(s)の部分はマネジメントコンソールでは省略して表示されており、クリックすると、詳細が表示されます。

time	value
2020-10-09 09:45:32.192000000	65.0
2020-10-09 11:16:40.034000000	45.0
2020-10-09 11:28:41.938000000	47.0
2020-10-09 11:33:45.047000000	55.0
2020-10-09 12:19:13.367000000	75.0
2020-10-09 12:36:46.970000000	10.0
2020-10-09 13:30:40.722000000	44.0
2020-10-09 13:56:28.837000000	60.0
2020-10-09 14:01:50.177000000	15.0
2020-10-09 14:03:38.020000000	0.0

Time series functions

時系列データに対する関数も用意されています。例えば、INTERPOLATE_LINEAR関数は値を埋めてくれます。

Fills in missing data using linear interpolation.

下の例は、30分ごとのデータを線形補間で生成しています。sequence(start, stop, step)関数で生成する値の間隔を指定していますが、startとstopは実際の値の範囲内じゃないとだめっぽいです。

SELECT truck_id, 
  INTERPOLATE_LINEAR(
    CREATE_TIME_SERIES(time, measure_value::double),
    SEQUENCE(min(time), max(time), 30m)
  ) as speed
FROM "IoT-Sample"."IoT" 
WHERE measure_name = 'speed'
GROUP BY truck_id
ORDER BY truck_id
LIMIT 10

truck_id	speed
1234546252	[ { time: 2020-10-09 09:45:32.192000000, value: 65.0 }, { time: 2020-10-09 10:15:32.192000000, value: 58.416049695656895 }... View 7 more row(s) ]
1575739296	[ { time: 2020-10-09 09:15:07.856000000, value: 56.0 }, { time: 2020-10-09 09:45:07.856000000, value: 45.53611215141335 }... View 8 more row(s) ]
1588092325	[ { time: 2020-10-09 09:01:49.081000000, value: 19.0 }, { time: 2020-10-09 09:31:49.081000000, value: 68.91556160771218 }... View 7 more row(s) ]
1641688615	[ { time: 2020-10-09 09:49:20.010000000, value: 37.0 }, { time: 2020-10-09 10:19:20.010000000, value: 47.34174652449723 }... View 7 more row(s) ]
1682738967	[ { time: 2020-10-09 09:07:58.814000000, value: 75.0 }, { time: 2020-10-09 09:37:58.814000000, value: 9.162118557623112 }... View 7 more row(s) ]
1712492054	[ { time: 2020-10-09 09:36:27.020000000, value: 44.0 }, { time: 2020-10-09 10:06:27.020000000, value: 34.61657724615213 }... View 6 more row(s) ]
1836816173	[ { time: 2020-10-09 08:59:28.330000000, value: 29.0 }, { time: 2020-10-09 09:29:28.330000000, value: 31.383254052802055 }... View 8 more row(s) ]
199744055	[ { time: 2020-10-09 09:16:35.398000000, value: 42.0 }, { time: 2020-10-09 09:46:35.398000000, value: 13.60871678326465 }... View 6 more row(s) ]
2062792987	[ { time: 2020-10-09 09:30:31.074000000, value: 45.0 }, { time: 2020-10-09 10:00:31.074000000, value: 52.07211022782296 }... View 7 more row(s) ]
21135517	[ { time: 2020-10-09 09:23:58.552000000, value: 64.0 }, { time: 2020-10-09 09:53:58.552000000, value: 57.93405176907693 }... View 8 more row(s) ]

データが30分おきになっています。

time	value
2020-10-09 09:45:32.192000000	65.0
2020-10-09 10:15:32.192000000	58.416049695656895
2020-10-09 10:45:32.192000000	51.83209939131379
2020-10-09 11:15:32.192000000	45.248149086970685
2020-10-09 11:45:32.192000000	60.18373944405349
2020-10-09 12:15:32.192000000	73.37867258972554
2020-10-09 12:45:32.192000000	15.522237945272241
2020-10-09 13:15:32.192000000	34.44762245218557
2020-10-09 13:45:32.192000000	53.213475743081105

min(time)からの30分間隔は少し気持ちが悪いので、00分、30分ごとにするにはこんな感じでしょうか

SELECT truck_id, 
  INTERPOLATE_LINEAR(
    CREATE_TIME_SERIES(time, measure_value::double),
    SEQUENCE(bin(min(time)+ 1h ,1h), max(time), 30m)
  ) as speed
FROM "IoT-Sample"."IoT" 
WHERE measure_name = 'speed'
GROUP BY truck_id
ORDER BY truck_id
LIMIT 10

truck_id	speed
1234546252	[ { time: 2020-10-09 10:00:00.000000000, value: 61.82577514127146 }, { time: 2020-10-09 10:30:00.000000000, value: 55.24182483692835 }... View 7 more row(s) ]
1575739296	[ { time: 2020-10-09 10:00:00.000000000, value: 40.34983728430808 }, { time: 2020-10-09 10:30:00.000000000, value: 29.77482942057512 }... View 6 more row(s) ]
1588092325	[ { time: 2020-10-09 10:00:00.000000000, value: 46.790964648508535 }, { time: 2020-10-09 10:30:00.000000000, value: 63.37907403508092 }... View 5 more row(s) ]
1641688615	[ { time: 2020-10-09 10:00:00.000000000, value: 39.1011371926136 }, { time: 2020-10-09 10:30:00.000000000, value: 59.68175770780711 }... View 6 more row(s) ]
1682738967	[ { time: 2020-10-09 10:00:00.000000000, value: 16.966425157908184 }, { time: 2020-10-09 10:30:00.000000000, value: 70.67966775160798 }... View 6 more row(s) ]
1712492054	[ { time: 2020-10-09 10:00:00.000000000, value: 45.89098423361806 }, { time: 2020-10-09 10:30:00.000000000, value: 12.212910765588772 }... View 5 more row(s) ]
1836816173	[ { time: 2020-10-09 09:00:00.000000000, value: 28.667924923980287 }, { time: 2020-10-09 09:30:00.000000000, value: 32.49950906732863 }... View 8 more row(s) ]
199744055	[ { time: 2020-10-09 10:00:00.000000000, value: 45.46459879696737 }, { time: 2020-10-09 10:30:00.000000000, value: 3.39022689038606 }... View 4 more row(s) ]
2062792987	[ { time: 2020-10-09 10:00:00.000000000, value: 50.54843613050327 }, { time: 2020-10-09 10:30:00.000000000, value: 21.4323793899003 }... View 6 more row(s) ]
21135517	[ { time: 2020-10-09 10:00:00.000000000, value: 56.71598240453767 }, { time: 2020-10-09 10:30:00.000000000, value: 50.6500341736146 }... View 6 more row(s) ]