Home/ Questions/Q 8569281
In Process

The Archive Base Latest Questions

Editorial Team
  • 0
Asked: 2026-06-11T18:17:53+00:00

I’ve read few articles regarding indexing in MongoDb, but have not got idea over

  • 0

I’ve read few articles regarding indexing in MongoDb, but have not got idea over physical layout of records. I got used to talk about clustered (quite fast physical based) and non-clustered indexes in relational database. There are no such terms for Mongo, though their doc mentioned secondary indexes. By default it seems to create index by _id primary key witch probably corresponds to physical order of item on a storage. Please explain me: if I create one index per table does it automatically store item in the physical order according to the index? If it is not, can I somehow set it up? what about _id, does it correspond physical order by default?

Share

Leave an answer

You must login to add an answer.

Need An Account,

1 Answer

  1. Editorial Team

    MongoDB indexes are B-tree indexes. Index blocks are allocated within the same datafiles that are used to store the documents. Currently (as of MongoDB version 2.2) there is no support for any other index type beyond the standard B-tree indexes.

    Ref: http://docs.mongodb.org/manual/core/indexes/

    MongoDB makes no attempt to order documents on disk, nor to place the B-Tree index blocks in any particular order. MongoDB uses memory-mapped files to access the on-disk data structures. As a result, the question of which index blocks are in RAM and which ones are paged out is delegated to the OS memory management system.

    Ref: http://docs.mongodb.org/manual/faq/storage/

    MongoDB documents are always contiguous on disk. Any document will only be in a single physical location: it is never necessary to assemble a document from multiple disk locations.

    MongoDB initially allocates documents on disk in the order in which they have been created. If a document grows beyond it’s allocated size (via updates to that document which add new fields, sub-documents, or array elements) then the document will be moved to a new location on disk, which is big enough to hold the new document.

    Deleting documents will create ‘holes’ in the allocated space: these holes are placed on a free list, and new documents get inserted into the these holes. As a result, if you perform repeated remove() and insert() operations on a MongoDB collection, the documents will be scattered across the disk in a highly non-ordered fashion.

    In particular, documents will NOT be laid out on the disk in _id order, or in the order of any other index.

    For more information about MongoDB storage management, take a look at these presentations:

    • 0