It is not true that a snapshot consumes less disk space than a disk backup.  It could actually consume more.  The space taken by the snapshot database is EXACTLY the same as the original database data files, minus the log file. 

If the database has a lot of available space, then the backup will consume less.

Snapshot database and Snapshot backups are two things.

Snapshot database is always <= Original database becasue it only contains the changed data.

Snapshot backup is more complex subject not direclty relevant to the database snapshot.

A snapshot database is NOT <= Original.  It will be = Original minus the log.

See proof below.  Commands in black, results in red.

exec sp_helpdb UPSRC   -- show size of original database (approx 301 GB)
 name            db_size       owner               
 --------------- ------------- --------------------
 UPSRC           301718.13 MB  RAINBIRD\PSAdmin    
 
 
 name            fileid filename                       size        
 --------------- ------ -----------------------------  -------------
 rbprod_data     1      F:\mssql\data\UPSRC_Data.mdf   253744832 KB
 rbprod_log      2      F:\mssql\data\UPSRC_Log.ldf    55214528 KB 

CREATE DATABASE UPSRC_Snap
ON (NAME = rbprod_data, FILENAME = 'f:\mssql\data\upsrc_snap.ss') AS SNAPSHOT OF UPSRC

exec sp_helpdb UPSRC_Snap  -- show size of snapshot database (approx 301GB)
 name           db_size      
 -------------  --------------
 UPSRC_Snap     301718.13 MB 
 
 
 name            fileid filename                       size         
 --------------- ------ ------------------------------ --------------
 rbprod_data     1      F:\mssql\data\UPSRC_Data.mdf   253744832 KB 
 rbprod_log      2      F:\mssql\data\UPSRC_Log.ldf    55214528 KB  

dir f:\mssql\data   (show size of database files on disk)
 
 11/27/2006  07:13 PM   259,834,707,968 UPSRC_Data.mdf
 11/27/2006  07:13 PM    56,539,676,672 UPSRC_Log.ldf
 12/11/2006  04:12 PM   259,834,707,968 upsrc_snap.ss

sp_helpdb shows you the maximum size of the file not the size actually used by the data in the files.

To find out the actual size of a sparse file

To learn the number of bytes each sparse file of the snapshot is currently using on disk, you can use the fn_virtualfilestats system table-valued function, which returns the current number of bytes in a file in the BytesOnDisk column. This function takes database_id and file_id as values. The database ID of the snapshot and the file IDs of each of its sparse files are displayed in the sys.master_files catalog view. For more information, see sys.master_files (Transact-SQL) and fn_virtualfilestats (Transact-SQL).

Alternatively, to see the disk space used by a sparse file, you can right-click the file in Microsoft Windows, click Properties, and look at the Size on disk value.

Bimal,

Thanks for the clarification.  I went so far as to examine the properties of the drive and found that after dropping the snapshot, that available space went up on the drive by exactly the amount shown in the "size on disk" property of the snapshot file.

This was very helpful.  I had abandoned use of the snapshot months ago when it looked like it was consuming the same space as the original.   We had noticed the performance hit in our testing and together with the perceived hit on disk space gave up. 

Dang...  I hate it when I am wrong.

Dave

Bimal,

Thanks for the clarification.  I went so far as to examine the properties of the drive and found that after dropping the snapshot, that available space went up on the drive by exactly the amount shown in the "size on disk" property of the snapshot file.

This was very helpful.  I had abandoned use of the snapshot months ago when it looked like it was consuming the same space as the original.   We had noticed the performance hit in our testing and together with the perceived hit on disk space gave up. 

Dang...  I hate it when I am wrong.

Dave