insertMany on M0 is extremely slow

Working with Data
,
#1

I have an insertMany with around 300 documents of an average and it takes on average 4 seconds. Is this due to my tier I am at at loss as to how I can improve the speeds here. I have reduced the document size as much as possible and it looks similar to this:

{
	"_id" : ObjectId("62a85e9017cfa90511b22c23"),
	"field" : {
		"field" : {
			"field" : 208.5,
			"field" : 208.5,
			"field" : 0
		},
		"field" : {
			"_id" : ObjectId("625fea09645b2228bcea11bc"),
			"field" : "text"
		},
		"field" : null,
		"field" : "text",
		"field" : "A8",
		"field" : "a8",
		"fields" : [ ],
		"field" : "text"
	},
	"field" : {
		"field" : {
			"key" : "text",
			"field" : "text",
			"field" : "text"
		},
		"_id" : ObjectId("61ba11a19d464624ae8f8a7d"),
		"field" : "text",
		"field" : "text",
		"field" : "text"
	},
	"field" : {
		"field" : {
			"field" : {
				"field" : 0,
				"field" : 0
			},
			"field" : 0,
			"fields" : [ ]
		},
		"fields" : 0
	},
	"field" : "text",
	"field" : ObjectId("62a76ea90ea8c7770e542a9e"),
	"field" : ObjectId("61ba0e5c63627829303591e0"),
	"__v" : 0,
	"field" : ISODate("2022-06-14T10:10:25.754Z"),
	"field" : ISODate("2022-06-14T10:10:25.754Z")
}

I changed names and values but just wanted to give an idea of its size. Is this normal? what recourse is there, please jebus tell me its a symptom of my tier :confused:

#2

By redacting you document to have only root field names, _id, __v and field, it make it totally unusable for us to experiment. Because, in JSON, where duplicated field names are replaced by the last occurrence of a repeated field you end up with the very small document:

{
  _id: ObjectId("62a85e9017cfa90511b22c23"),
  field: 2022-06-14T10:10:25.754Z,
  __v: 0
}

M0 is shared so the performance you experiment is dependant of what else is happening of the shared server.

What is the average size of your 300 documents?

What are the total size of your 300 documents?

What is your connection between your client and the rest of internet?

Where is your location compared to the region of the cluster?

All the above influence the performances.

One test at 4 seconds might be normal but if you consistently clock at 4s that seems slow. If you are doing load testing on a shared server you are shooting your self and everybody else in the foot.

#4

Hi in your experience does upgrading to a dedicated tier on average improve insert performance

#5

Dedicated will always be better than shared. And more predictable as all the traffic will be yours. Your performances will not be influence by the other applications using the same shared server.

You have not answered any of my questions, so nobody can know for sure if you will gain from the switch. Your bottleneck has not been determined yet.