Question

I have a collection in MongoDB with 20 million documents and an index created for the collection. I have confirmed that the find() query uses the index (col_1_col_2) (confirmed using the explain). However, I’ve noticed that the very first find() query in a day takes 5-10 sec, but all subsequent queries take less than 100 ms.

Updated DB stats:
db..stats()
ns .
count 20135438
size 5694.57m
avgObjSize 296.55146314672
storageSize 7109.57m
numExtents 35
nindexes 4
lastExtentSize 1188.86m
paddingFactor 1
flags 1
totalIndexSize 3044.69m
indexSizes {
“id“: 653319632,
“col_1_col_2”: 691730480,
“col2_1”: 848832320,
“col1_1”: 998706576
}
ok 1
{top:1}
total 18
readLock 16

Questions:
1 Why does the first query take longer time?
2 Is this to do with loading the index in RAM?

Answer 1

check the disk activity during this first request. It could be related to mounting data from disk to RAM.

If you use linux : iostat -m 1 will help you.