File system snapshot

Creating a backup of a large amount of data can take a long time. In multi-tasking or multi-user systems , files or directories may be recorded or modified during backups, which can lead to incorrect data backups. For example, a user moves a file from a directory that was not yet saved during the backup (backup - backup), to a directory that has already been saved. Such a file may not be included in the backup at all. Also, a file intended for backup may be written at the time of reading by the backup procedure and may be saved in the wrong version.

One of the methods to safely back up is to prohibit writing to the data to be backed up while the backup is being created. Another method is to stop all applications that can change this data, or to block these applications by forcing read-only mode using the application programming interface (API) of the operating system (OS). These methods are used in low-availability systems (home computers, small workgroup servers for which regular downtime is allowed). In 24/7 high-availability systems, these methods cannot be used, as this may result in denial of service.

To avoid downtime, high-availability systems can instead of direct backup first create a snapshot - a read-only copy of information that is “frozen” at a certain point in time. And then, allowing applications to continue updating data, back up. Most snapshot implementations are effective; they create a snapshot for O (1) . In other words, the time and number of I / O operations required to create a snapshot does not increase with the growth of the data volume, while the same parameters for creating a direct backup are proportional to the size of the data being saved.

Read-write snapshots sometimes cause branching of snapshots, as they implicitly create different versions of their data. In addition to backing up and restoring data, snapshots are often used in virtualization, in various sandboxes and in shared hosting due to their effectiveness in managing changes to a large data set.

Volume Management

Some operating systems from the UNIX family, for example, HP-UX , as well as Linux, have a logical partition manager that supports snapshots. This implementation of copying when recording on entire block devices by copying the changed blocks, before they are overwritten, to another place, preserves the integrity of the snapshot on the block device. File systems on this snapshot can be later connected only in read mode. Block-level snapshots are almost always less efficient in using space than snapshots in file systems that support them.

File Systems

Some file systems, such as WAFL , fossil for Plan 9, or ODS-5 , internally track old versions of files and make snapshots available through a special namespace . Others, such as UFS2 , provide APIs for the operating system to access their file history. NTFS provides access to snapshots through Volume Shadow Copy (VSS) on Windows XP and Windows Server 2003 , and through Shadow Copy on Windows Vista . Snapshots are also available on Novell Storage Services (NSS), the file system for Netware , starting with version 4.11, and newer on Linux platforms in Open Enterprise Server (OES) products.

ZFS has a hybrid implementation that tracks read / write snapshots at the block level, but creates branched file sets known to users as “clones”.

In Databases

The SQL specification defines four levels of transaction isolation . At the highest level - serialization , a snapshot is created at the start of each transaction . Utilities for backing up most popular SQL databases use this technology to create a self-consistent image of a data table.

Other applications

Software transactional memory is a scheme that applies a similar concept to data structures stored in RAM.

• Check Point
• Backup
• Time Machine (program)
• File system list
• File System Comparison

• Mikhail Mitroshin. Problems, What is snapshot? / BrightStor ARCserve Backup - snapshot technologies (Russian) . Storage News (No. 4 (25) 2005). Date of treatment August 3, 2010. Archived March 15, 2012.
• Neeta Garimella. Understanding and exploiting snapshot technology for data protection, Part 1: Snapshot technology overview (neopr.) . IBM (04/26/2006). Date of treatment August 3, 2010. Archived March 15, 2012.
• Mike Harwood. Storage Basics: Backup Strategies (unopened) (link not available) . Storage (09.24.2003). Date of treatment August 3, 2010. Archived on April 16, 2008.