MongoDB数据库的索引操作（转）

时间：2010-10-21 来源：不得闲

> use user
switched to db user
> show collections
system.indexes
u_info
u_setting
> db.system.indexes.find(); 这是默认的索引（默认为_id为索引）
{ "name" : "_id_", "ns" : "user.u_info", "key" : { "_id" : ObjectId("000000000000000000000000") } }
{ "name" : "_id_", "ns" : "user.u_setting", "key" : { "_id" : ObjectId("000000000000000000000000") } }
>
> db.u_info.insert({uid:1,name:"Falcon.C",address:"Beijing"});
> db.u_info.insert({uid:2,name:"sexMan",address:"Wuhan"});
> db.u_info.find();
{ "_id" : ObjectId("4b9cf280c84d7f20576c4df2"), "uid" : 1, "name" : "Falcon.C", "address" : "Beijing" }
{ "_id" : ObjectId("4b9cf284c84d7f20576c4df3"), "uid" : 2, "name" : "sexMan", "address" : "Wuhan" }

插入了2条记录，我们来把uid设置为索引字段：

> db.u_info.ensureIndex({uid:1});
> db.u_info.ensureIndex({name:1});
> db.system.indexes.find();
{ "name" : "_id_", "ns" : "user.u_info", "key" : { "_id" : ObjectId("000000000000000000000000") } }
{ "name" : "_id_", "ns" : "user.u_setting", "key" : { "_id" : ObjectId("000000000000000000000000") } }
{ "ns" : "user.u_info", "key" : { "uid" : 1 }, "name" : "uid_1" }
{ "ns" : "user.u_info", "key" : { "name" : 1 }, "name" : "name_1" }
>

这时我们看到多了刚才我们设置的那个字段，这样在查询的时候，如果查询条件有uid字段或name字段，则走索引来进行查询

有索引：

> db.u_info.find({name:"Falcon.C"});
{ "_id" : ObjectId("4b9cf280c84d7f20576c4df2"), "uid" : 1, "name" : "Falcon.C", "address" : "Beijing" }
> db.u_info.find({name:"Falcon.C"}).explain();
{
      "cursor" : "BtreeCursor name_1",
      "startKey" : {
               "name" : "Falcon.C"
      },
      "endKey" : {
               "name" : "Falcon.C"
      },
      "nscanned" : 1,
      "n" : 1,
      "millis" : 0,
      "allPlans" : [
               {
                     "cursor" : "BtreeCursor name_1",
                     "startKey" : {
                              "name" : "Falcon.C"
                     },
                     "endKey" : {
                              "name" : "Falcon.C"
                     }
               }
      ]
}

删除索引后：

> db.system.indexes.find();
{ "name" : "_id_", "ns" : "user.u_info", "key" : { "_id" : ObjectId("000000000000000000000000") } }
{ "name" : "_id_", "ns" : "user.u_setting", "key" : { "_id" : ObjectId("000000000000000000000000") } }
{ "ns" : "user.u_info", "key" : { "uid" : 1 }, "name" : "uid_1" }
{ "ns" : "user.u_info", "key" : { "name" : 1 }, "name" : "name_1" }
> db.u_info.dropIndex("name_1")
{ "nIndexesWas" : 3, "ok" : 1 }
> db.u_info.find({name:"Falcon.C"}).explain();
{
      "cursor" : "BasicCursor",
      "startKey" : {

      },
      "endKey" : {

      },
      "nscanned" : 2,
      "n" : 1,
      "millis" : 0,
      "allPlans" : [
               {
                     "cursor" : "BasicCursor",
                     "startKey" : {

                     },
                     "endKey" : {

                     }
               }
      ]
}
> db.system.indexes.find();
{ "name" : "_id_", "ns" : "user.u_info", "key" : { "_id" : ObjectId("000000000000000000000000") } }
{ "name" : "_id_", "ns" : "user.u_setting", "key" : { "_id" : ObjectId("000000000000000000000000") } }
{ "ns" : "user.u_info", "key" : { "uid" : 1 }, "name" : "uid_1" }

通过以上可以看出，查询的条件中有索引时，查询走BtreeCursor 的索引，而没有索引时走BasicCursor

通常需要索引的字段是：
1.唯一键 _id 是默认被设置为索引的
2.需要被查找的字段应该建立索引,比如在find()里面的字段
3.需要被排序的字段应该建立索引。比如在sort()里面的字段

以上就是MongoDB

的索引操作