Search Current Space

compared with
Current by Robert Crockett
on Nov 01, 2012 09:24.

This line was removed.
This word was removed. This word was added.
This line was added.

Changes (3)

View Page History
Block based backups bypass files and file systems almost completely. All operating systems have a specialized component of the O/S called the [File System|]. Examples of popular server file systems are [NTFS |]on Windows and [ext3|] for Linux. The file system divides the hard disk, volume or RAID array (software and hardware RAID) into chunks or groups of bytes called blocks (fixed size) or extents (variable size).  Typically these are ordered 0 - N. The size of a block depends on the file system used and potentially settings used when the file system was created. NTFS and ext3 use fixed block sizes.  Some file systems support a concept of variable length block sizes typically called extents.

The file system is responsible for keeping track of the tree or hierarchy of files. It also stores a file in neat little fixed size blocks on the disk and keeps track of where these blocks are which can be scattered across the disk.  Backup applications that read files use the file system to get at data and are inherently very slow and time consuming no matter what file system is used.
Some backup application vendors use methods for taking snapshots that do not guarantee file system consistency. These should be avoided when possible.  In the case of 2.4 Linux it is unavoidable.  Linux kernels greater than 2.6.8 have low level device driver facilities for making file systems consistent during a snapshot.  Windows XP and greater support Volume Shadow COpy which gauruntees a consistent file system during a snapshot.  Beware of applications that brag about their method for determing a snapshot using a "quiesce  period".  You can identify these by reading about their snapshot process.  They will usually describe a process that waits for a quiesce period (e.g. 5 seconds) to be "observed" and once the disk is quiesced or observed to be quiet it is therefore "assumed" the file system is in a consistent state\!  For an example see the OTM technical paper at []

R1Soft takes a different approach.  R1Soft has made a comitment to being cross-platform on Windows and Linux.  Truely cross platform.  This means you can use a Linux CDP Server to perform Continuous Data Protection for both Windows and Linux and the opposite is also true.  You can use a Windows CDP Server to perform Continuous Data Protection for both Linux and Windows.  For example this means you you prefer to maintain Windows servers and have to provide backup services for Linux servers you can do it all with Windows.

R1Soft is the only vendor doing and we do it using special proprietary file system code that can read files and file attributes from raw disk blocks in a platform independent fashion.  This means we don't depend on the O/S when we read files.  We do it ourselves.  And this has some big advantages.  In addition to being cross platform in how you manage your server backups.  It allows is to do cross-platform file restores.  You can restore files out of block-based backup done on a Linux server to a Windows server and visa-versa.  R1Soft can also transfer the files for you at restore time.  The other applications require that you setup your own NFS or CIFS share for the backup application to store and read backup data from.  R1Soft has its own network protocol for moving the files and block data so you don't have configure a network file system.