是否有可能在MongoDB中反转索引的遍历顺序以保持性能但是创建类似的索引？

Question

集合c包含数十万个文档，每个文档都有两个字段（x和y）。以下是一些例子。

{ "_id" : ObjectId("53fecab45ae4ec5280a0736f"), "x" : 1, "y" : 1 }
{ "_id" : ObjectId("53fecab45ae4ec5280a07370"), "x" : 2, "y" : 1 }
{ "_id" : ObjectId("53fecab45ae4ec5280a07371"), "x" : 3, "y" : 1 }
{ "_id" : ObjectId("53fecab45ae4ec5280a07372"), "x" : 3, "y" : 2 }
{ "_id" : ObjectId("53fecab45ae4ec5280a07373"), "x" : 3, "y" : 3 }
{ "_id" : ObjectId("53fecab45ae4ec5280a07374"), "x" : 4, "y" : 2 }
{ "_id" : ObjectId("53fecab45ae4ec5280a07376"), "x" : 5, "y" : 1 }
{ "_id" : ObjectId("53fecab45ae4ec5280a07377"), "x" : 5, "y" : 3 }
{ "_id" : ObjectId("53fecab45ae4ec5280a07378"), "x" : 5, "y" : 5 }
{ "_id" : ObjectId("53fecab45ae4ec5280a07379"), "x" : 6, "y" : 2 }
{ "_id" : ObjectId("53fecab45ae4ec5280a0737a"), "x" : 6, "y" : 4 }
{ "_id" : ObjectId("53fecab45ae4ec5280a0737b"), "x" : 6, "y" : 4 }
{ "_id" : ObjectId("53fecab45ae4ec5280a0737c"), "x" : 7, "y" : 3 }
{ "_id" : ObjectId("53fecab45ae4ec5280a0737d"), "x" : 8, "y" : 8 }
{ "_id" : ObjectId("53fecab45ae4ec5280a0737e"), "x" : 9, "y" : 3 }
{ "_id" : ObjectId("53fecab45ae4ec5280a0737f"), "x" : 9, "y" : 1 }
{ "_id" : ObjectId("53fecab45ae4ec5280a07380"), "x" : 10, "y" : 11 }

x和y的值是数字类型，但是随机。 现在，我想找到不超过3个文档，其值为least但larger than or equal to {x：5，y：3} 。 按功能比较两份文件的规定：

int compareDocument(doc1, doc2) {
    var result;
    if (doc1.x > doc2.x) {
        result = 1;
    } else if (doc1.x == doc2.x) {
        if (doc1.y > doc2.y) {
            result = 1;
        } else if (doc1.y == doc2.y) {
            result = 0;
        } else {
            result = -1;
        }
    } else {
        result = -1;
    }
    return result;
}

必须以高性能执行。我的方法是：

> db.c.ensureIndex({x:1,y:1},{name:"asc"});
> db.c.find().hint("asc").min({x:5,y:3}).limit(3);
{ "_id" : ObjectId("53fecab45ae4ec5280a07377"), "x" : 5, "y" : 3 }
{ "_id" : ObjectId("53fecab45ae4ec5280a07378"), "x" : 5, "y" : 5 }
{ "_id" : ObjectId("53fecab45ae4ec5280a07379"), "x" : 6, "y" : 2 }
> db.c.find().hint("asc").min({x:5,y:3}).limit(3).explain();
{
        "cursor" : "BtreeCursor asc",
        "isMultiKey" : false,
        "n" : 3,
        "nscannedObjects" : 3,
        "nscanned" : 3,
        "nscannedObjectsAllPlans" : 3,
        "nscannedAllPlans" : 3,
        "scanAndOrder" : false,
        "indexOnly" : false,
        "nYields" : 0,
        "nChunkSkips" : 0,
        "millis" : 0,
        "indexBounds" : {
                "start" : {
                        "x" : 5,
                        "y" : 3
                },
                "end" : {
                        "x" : {
                                "$maxElement" : 1
                        },
                        "y" : {
                                "$maxElement" : 1
                        }
                }
        },
        "server" : "Mars-PC:27017"
}

以上仅扫描了3份文件。

同样，我想找到不超过3个文件，其值为largest，但less than or equal to {x：5，y：3} 。 是否可以通过使用上述索引asc或其他内容而不创建新索引来实现此目的，但保持高性能？

我找不到它。我能做的是创建另一个类似的索引来实现它。

> db.c.ensureIndex({x:-1, y:-1},{name:"desc"});
> db.c.find().hint("desc").min({x:5, y:3}).limit(3);
{ "_id" : ObjectId("53fecab45ae4ec5280a07377"), "x" : 5, "y" : 3 }
{ "_id" : ObjectId("53fecab45ae4ec5280a07376"), "x" : 5, "y" : 1 }
{ "_id" : ObjectId("53fecab45ae4ec5280a07374"), "x" : 4, "y" : 2 }
> db.c.find().hint("desc").min({x:5, y:3}).limit(3).explain();
{
        "cursor" : "BtreeCursor desc",
        "isMultiKey" : false,
        "n" : 3,
        "nscannedObjects" : 3,
        "nscanned" : 3,
        "nscannedObjectsAllPlans" : 3,
        "nscannedAllPlans" : 3,
        "scanAndOrder" : false,
        "indexOnly" : false,
        "nYields" : 0,
        "nChunkSkips" : 0,
        "millis" : 0,
        "indexBounds" : {
                "start" : {
                        "x" : 5,
                        "y" : 3
                },
                "end" : {
                        "x" : {
                                "$minElement" : 1
                        },
                        "y" : {
                                "$minElement" : 1
                        }
                }
        },
        "server" : "Mars-PC:27017"
}