Why $in is much faster then $all

Question

I am well aware of what those 2 do. $in take .3 seconds. And $all take 26 seconds. Huge different. If anything $in should take longer because it returns more data.

Also if the limit is removed, $in actually is much faster.

db.tablebusiness.find({ "LongitudeLatitude" : { "$nearSphere" : [106.772835, -6.186753], "$maxDistance" : 0.053980478460939611 }, "Prominent" : { "$gte" : 15 }, "indexContents" : { "$all" : [/^soto/, /^nasi/] } }).limit(200);

db.tablebusiness.find({ "LongitudeLatitude" : { "$nearSphere" : [106.772835, -6.186753], "$maxDistance" : 0.053980478460939611 }, "Prominent" : { "$gte" : 15 }, "indexContents" : { "$in" : [/^soto/, /^nasi/] } }).limit(200);

This is result

/* 88 */
{
  "ts" : ISODate("2012-09-11T06:57:26.801Z"),
  "op" : "query",
  "ns" : "newisikota.tablebusiness",
  "query" : {
    "LongitudeLatitude" : {
      "$nearSphere" : [106.772835, -6.186753],
      "$maxDistance" : 0.053980478460939611
    },
    "Prominent" : {
      "$gte" : 15.0
    },
    "indexContents" : {
      "$all" : [/^soto/, /^nasi/]
    }
  },
  "ntoreturn" : 200,
  "nscanned" : 48,
  "nreturned" : 48,
  "responseLength" : 60002,
  "millis" : 3821,
  "client" : "127.0.0.1",
  "user" : ""
}

/* 89 */
{
  "ts" : ISODate("2012-09-11T06:57:43.147Z"),
  "op" : "query",
  "ns" : "newisikota.tablebusiness",
  "query" : {
    "LongitudeLatitude" : {
      "$nearSphere" : [106.772835, -6.186753],
      "$maxDistance" : 0.053980478460939611
    },
    "Prominent" : {
      "$gte" : 15.0
    },
    "indexContents" : {
      "$in" : [/^soto/, /^nasi/]
    }
  },
  "ntoreturn" : 200,
  "nscanned" : 200,
  "nreturned" : 200,
  "responseLength" : 249598,
  "millis" : 320,
  "client" : "127.0.0.1",
  "user" : ""
}

Note: the $all query can run for 26 seconds once in a while.

The explain result is the following:

db.tablebusiness.find({ "LongitudeLatitude" : { "$nearSphere" : [106.772835, -6.186753], "$maxDistance" : 0.053980478460939611 }, "Prominent" : { "$gte" : 15 }, "indexContents" : { "$all" : [/^soto/, /^nasi/] } }).limit(200).explain();




{
        "cursor" : "GeoSearchCursor",
        "nscanned" : 48,
        **"nscannedObjects" : 48,**
        "n" : 48,
        "millis" : 8563,
        "nYields" : 0,
        "nChunkSkips" : 0,
        "isMultiKey" : false,
        "indexOnly" : false,
        "indexBounds" : {
        }
}
>


db.tablebusiness.find({ "LongitudeLatitude" : { "$nearSphere" : [106.772835, -6.186753], "$maxDistance" : 0.053980478460939611 }, "Prominent" : { "$gte" : 15 }, "indexContents" : { "$in" : [/^soto/, /^nasi/] } }).limit(200).explain();
{
        "cursor" : "GeoSearchCursor",
        "nscanned" : 200,
        **"nscannedObjects" : 200,**
        "n" : 200,
        "millis" : 516,
        "nYields" : 0,
        "nChunkSkips" : 0,
        "isMultiKey" : false,
        "indexOnly" : false,
        "indexBounds" : {
        }
}

Notice that the $in search scan more object.

Answer 1

$in can abort and return success as soon as the first match is found, while $all has to go through all the data.

The runtime is dominated by the time needed to check for matches, not by the time needed to return data, because any data that does get returned has just been checked for a match and is already in the cache.