MongoDB Pro Hint — Use ObjectID as Timestamp

Did you know MongoDB’s ObjectID is actually a timestamp?
It is created by getting the timestamp at the creation time of a document, a unique machine indicator, and a random number. The combination of those parts produces a 12-byte sequence, 24 hexadecimal chars (e.g: “62bdb3978381b52a31ea0a07”).

When a new collection is created, MongoDB creates a default index on _id field.
By utilizing this fact, one can easily query a large DB collection without any relevant indexes, or indexes at all, relatively efficiently.

For example, you have a collection myCollection, which contains 100M documents. Not indexes at all, or non that are relevant to what you search.
If you write your query as:

db.myCollection.find({ myField: { $lt: 10 } });

Not having relevant indexes will force the DB to perform a Full Collection Scan (COLLSCAN) and will result in a slow-running query.
By performing this small adjustment you’ll cause the DB to utilize the default index of _id and cut the working set, so any of your non-indexed fields will cause smaller set to search within:

db.myCollection.find({ _id: { $gt: ObjectID(“62bdb3978381b52a31ea0a07”) }, myField: { $lt: 10 } });

Or:

db.myCollection.find({
_id: {
$gte: ObjectID(“62bdb13b0000000000000000”) ,
$lt: ObjectID(“62bc5fbb0000000000000000”) ,
}, myField: { $lt: 10 } });

This will cut the dataset for a single day time-frame and then perform the rest of the non-indexed document search.
Notice that the ObjectIDs above are with zeros. That’s because we only care about the timestamp portion of the ObjectID and not about the other parts. And since timestamps are essentially numeric values and sequential, this allows us to sort by them.

Obviously, this is only applicable when you have a known time frame.

Here is a small utility to easily create and parse ObjectIDs to Timestamp and vice versa:

https://liv.tools/utilities/objectId/62bdb3978381b52a31ea0a07