MongoDB sensors storage strategy - database

Question

I know that this is a broad questions so I'll try to set some boundaries to the domain of this question:
Let's assume that the data can arrive at any time from a source and the period is not fixed between data acquisition.
The data is represented using a generic index (time-based, length-based or index based)
The data payload could miss some "columns" (if we imagine a data chunk as a table)
An example of this scenario could be the following:
{ │ {
"timestamp": t3, │ "timestamp": t1, # t1 < t3
"temperature": 40, │ "temperature": 40,
┌───────────┐ "pressure": 1220 │ "color": "ffffff" ┌───────────┐
│ │ } │ } │ │
│ │ │ │
│ Source 1 ├─────────────────────────────────────────────────────────►│ │
│ │ │ │
│ │ │ │
└───────────┘ │ │
│ │ │
{ │ { │ │
"length_meters": l1,│ "length_meters": l2, │ │
┌───────────┐ "resistivity": 40 │ "resistivity": 40 │ │
│ │ } │ } │ │
│ │ │ MongoDB │
│ Source 2 ├─────────────────────────────────────────────────────────►│ │
│ │ │ │
│ │ │ │
└───────────┘ │ │
│ │
│ │
{ │ │
"index": 1, │ │
┌───────────┐ "description: "lorem" │ │
│ │ } │ │
│ │ │ │
│ Source 3 ├─────────────────────────────────────────────────────────►│ │
│ │ │ │
│ │ └───────────┘
└───────────┘
I've read the official documentation on the Bucket pattern in which they describe this technique which is, in my opinion, far from reality since in the example they suppose to store a single sensor data (temperature, in that case). Furthermore I don't like the solution since is repeating the column (temperature) for every sample of the sensors.
Is there any better and generic way to design the collection(s) to tackle such requirements which uses contained storage size and can enable good performance in terms of data querying?
Samples queries could be the following:
Return pressure and temperature between tx and ty.
Return the maximum temperature between tx and ty.
Return the minimum of resistivity between lx and ly.

Answer 1

Based on the sample data, the buckets could be also like this:
{
sensor_id: 12345,
start_date: ISODate("2019-01-31T10:00:00.000Z"),
end_date: ISODate("2019-01-31T10:59:59.000Z"),
timestamp: [
ISODate("2019-01-31T10:00:00.000Z"),
ISODate("2019-01-31T10:01:00.000Z"),
ISODate("2019-01-31T10:42:00.000Z")
],
temperature: [
40,
40,
42,
]
}
What do you mean by "good performance in terms of data querying"? It heavily depends ont the queries you like to run.