Backup system

Structure of system operation of backup

Backup systems provide the continuity of business processes and data protection from natural and technogenic catastrophes, actions of malefactors. These technologies are actively used in IT infrastructures of the organizations of the most different industries and scales.

Data backup — process of creation of the copy of data on the carrier intended for data recovery in the original place of their arrangement in case of their damage or destruction. Besides, the backup system is one of necessary methods of ensuring business continuity. Creation of a centralized system of backup allows to reduce total cost of ownership IT infrastructure thanks to optimal use of backup devices and cost reduction by administration (in comparison with hostless system).

Organizational difficulties in data protection

• Internal contradictions in a technical command
• Administrators of applications should be responsible for safety of data, SLA and recovery?
• The centralized automated control – risk reduction for the Chief information officer: the transparency, predictability of IT of processes increases

Correct strategy of data protection for DPC

Outdated approach under the name "BACKUP" ^[1]

• Backup
• Recovery

Modern approach under the name "INFORMATION MANAGEMENT"

• Backup
• Recovery
• Analytics on contents
• Context search
• Mobile data access
• Transparent integration with a cloud
• Problems of cybersecurity
• ANY applications of third-party developers on data processing (Open API)

Problem of copies

• In the absence of the centralized approach the number of data uncontrollable grows
• Where does the most current version of data lie?
• If it is required to delete data on Compliance, where to find all copies?
• Udaleniye archiving of outdated information. How to define reasonable criterion of value of data?

Architecture and system operation of backup

The centralized system of backup has a multi-tier architecture which enter:

• the server of management of backup capable also to combine functions of the server of copying of data;
• one or several servers of copying of data, to which backup attached devices;
• computers clients with the programs agents of backup set on them;
• console of the system administrator of backup.

The system administrator keeps the list of computers clients of backup, recording devices and carriers of storage of reserve data and also makes the backup schedule. All this information contains in special base which is stored on the server of management of backup.

According to the schedule or at the command of the operator the server of management gives a command to the program agent set on the computer client to begin data backup according to the selected policy. The program agent collects and transfers data which are subject to reservation to the copying server specified by it the server of management.

The server of copying saves data retrieveds on the data storage device connected to it. Information on process (what files were copied on what carriers copying, etc. was performed) remains in base of the server of management. This information allows to find location of the saved data in need of their recovery on the computer client.

That the backup system saved consistent data of the computer client, they should not be modified in the course of their collecting and copying by the program agent. For this application of the computer client should complete all transactions, save contents a cache memory on a disk and suspend the work. This process is initiated at the command of the program agent which is transferred to applications of the computer client.

As the backup system is intended for data recovery after failure or accidents, the created backup copies need to be checked regarding integrity and working capacity. Besides, at creation of a backup system it is necessary to keep within the reduced backup "window". Generally speaking, the requirement of round-the-clock operation of information systems reduces practically to zero the available time frame of a stop of applications necessary for implementation of backup operation (backup "window").

Backup classification

On completeness of the saved information

• Full redundancy (Full backup) — creation of reserve archive of all system files, usually including system status, the register and the other information necessary for complete recovery of workstations. So not only files, but also all information necessary for system operation are reserved.
• Additional reservation (Incremental backup) — creation of reserve archive from all files which were modified after the previous full or additional redundancy.
• Differential reservation (Differential backup) — creation of reserve archive from all files which were changed after the previous full redundancy.
• Selective reservation (Selective backup) — creation of reserve archive only from the selected files.

On a method of access to the carrier

• Operational reservation (Online backup) — creation of reserve archive on constantly connected (directly or through network) the carrier.
• Autonomous reservation (Offline backup) — storage of the backup copy on the removable medium, the cartridge or a cartridge which before use should be installed in the drive.

Rules of work with backup systems

When using any technology of backup it is necessary to follow some fundamental rules which observance will ensure the maximum safety of the unforeseen situations given in case of emergence.

• Preliminary planning. In planning process all components of infrastructure of backup should be considered, and all applications, servers and trends of increase in capacity of primary data warehouses should not remain unaddressed.
• Establishment of lifecycle and calendar of transactions. All tasks connected with backup should be documented and be executed according to the schedule. Below the task list which need to be executed daily is provided:
  • monitoring of tasks;
  • reports on failures and successful accomplishment;
  • analysis and problem resolution;
  • manipulations with tapes and management of library;
  • drawing up schedule of task performance.

• Daily overview of logs of process of backup. As each failure in creation of backup copies can cause a set of difficulties, it is necessary to check the copying process course, at least, an every day.
• Protection of the database of backup or directory. Each backup application maintains the database which loss can mean loss of backup copies.
• Daily determination of a temporary window of backup. If runtime of tasks begins to go beyond the taken-away temporary window, it is sign of approach to the extreme capacity of a system or existence of weak links in performance. Timely detection of such signs can save from the subsequent larger failures of a system.
• Localization and preserving of the "external" systems and volumes. It is necessary to check personally compliance of backup copies to the expectations, first of all relying on an owl of observation, but not on reports of programs.
• Greatest possible centralization and automation of backup. Data of a set of tasks of reservation in one considerably simplify process of creation of copies.
• Creation and support of open reports, reports on open problems. Existence of the magazine of unresolved problems can promote their fastest elimination, and, as a result, backup process optimization.
• Inclusion of backup in process of change control of a system.
• Consultations with vendors. It is necessary to be convinced that the implemented system completely meets expectations of the organization.

Technologies of backup

Neither RAID, nor cluster, nor any other technology of ensuring fault tolerance protect from errors as a result of which data change or removed and of which the operating system or the person are guilty. Backup — one of optimal solutions for such situations as it allows to store copies of different prescriptive limit, for example for each day of the current week, two-week, monthly, semi-annual and annual prescription. An opportunity to use external removable mediums significantly reduces information storage costs, however alternative technologies are suitable for some tasks more.

Backup using SAN

Application the Storage Area Network allows to transfer completely backup traffic from a local network to network of storage. There are two options of implementation: without loading of a local network, or off-network copying (LAN-free backup), and without participation of the server, or extra server copying (Server-free backup).

Off-network copying

At off-network copying data from a disk are also back transferred to a tape in SAN. The exception of a network segment of a way of data backup allows to avoid excessive delays on transfer of traffic through IP network and I/O cards. Loading of a local network falls, and backup can be carried out practically at any time. However transfer of data is executed by the server connected to SAN that increases load of it. Thanks to the Fibre Channel protocol using one optical cable several data transmission channels can be organized. At the same time all volume of the reserved data from backup-servers of storage is sent to the tape device, passing a local network. In this case the local network is necessary only for control of operation of backup-servers from master servers. Thus, only the small volume of metadata which contain information on the reserved data, is transferred on a local network. Master servers are responsible in general for policy of data backup in the segment or an area of responsibility. All backup-servers in relation to a master server are clients. It is considered that the considered method of backup can involve the peak bandwidth of the Fibre Channel as much as possible.

In quality protocol, applied to data transmission between servers and libraries, can be used both SCSI atop Fibre Channel, and IP atop Fibre Channel especially as the majority FC- adapters and FC- hubs work along with both IP SCSI protocols (i) at one Fibre Channel- the channel.

Extra server copying

This type of backup represents further development of a method of off-network copying (LAN-free) as reduces the number of processors, memories, input/output devices involved in this process. This process archives sections entirely, unlike pofaylovy archiving, but at the same time allows to recover separate files. By determination, at extra-server copying data are copied from a disk on a tape and back without direct participation of the server. As backup requires existence of some additional third node which is completely responsible for copying process, from there is also other name of this approach — copying to participation of the third party (Third_-Party Copy, 3PC). So, as the similar equipment the router of data warehouses which undertakes the functions which were earlier executed by the server can be used.

One of advantages of architecture of SAN — lack of a tough binding of systems making it to any data storage devices. This property is also put in fundamentals of technology of backup without participation of the server. In this case to the disk array can have direct access both the server of data, and the devices which are taking part in copying from disk arrays. Backup of the data units relating to any file is preceded by creation of a certain index or list of numbers of the blocks belonging to it. It also allows to involve external devices to backup further.

Thus, extra server copying allows to move directly data between the disk arrays and libraries connected to SAN network. At the same time data move on SAN network and do not load either a local network, or servers. Such copying is considered ideal for corporate networks which should function in a continuous duty 24 hours a day, 7 days a week. Especially for for what the time period during which it is possible to execute backup without significant effect on work of users and applications becomes unacceptably small.

Replication of data

Modern disk arrays have tools for copies of data in the array. The data created by these means carry the name Point-In-Time (PIT) - copies, i.e. fixed at a certain point in time. There are two types of tools for PIT copies: cloning and "instant picture" (snapshot). Usually understand complete copying of data as cloning. It requires as much disk space, as well as for initial data, and some time. When using such copy there is no load of the disk volumes containing initial data. In other words, there is no additional load of a disk subsystem of the productive server.

The work mechanism of "instant pictures" other can be also implemented both programmatically on the productive server, and hardware in an array. At the moment when it is necessary to begin backup, the program agent gives a command to the application to complete all transactions and to save a cache memory on a disk. Then the virtual structure — snapshot representing the card of arrangement of data units which OS and another software perceives as a logical volume is created. The application interrupts a standard mode of work for a short time, necessary for preserving of data. After that the application continues to work in a standard mode and to change data units, at the same time before change old data of the block using the snapshot driver are copied to the area a cache memory of snapshot and in the card of arrangement of data units the link to new location of the block is specified. Thus, the snapshot card always indicates the data units received at the time of completion of transactions by the application. Data units which were not changed are stored on the former place, and old data of the changed blocks — in area snapshot cache memory. The program agent copies the consistent data obtained at the time of completion of transactions by the application, providing access to them via the snapshot driver, i.e. using the card of arrangement of blocks. Creation of copies using "instant pictures" saves disk space, but creates additional load of a disk subsystem of the productive server. What of methods of creation of PIT copies to select, is solved at a design stage of a backup system, proceeding from the business requirements shown to a system.

Security

As a rule, backup is automatic. For data access the increased privileges usually are required. So the process providing backup is started from under an account with the increased privileges — here and a certain risk creeps in. Read article the Backup system (security).

World market of backup systems

According to the analysis which is carried out by analytical company IDC in an early autumn of 2008 the market (however as well as the majority of spheres and services in the field of IT) considerably grew, the benefit as well as profits of the companies specializing in production of means data protection. Read article"Data backup (world market)"

Russian market of backup systems

Read article"Data backup (market of Russia)"

Examples of backup systems

The complete directory of backup systems and implementation projects is available on TAdviser.

You See Also

• Storage systems
• Direct Attached Storage, DAS
• NAS
• Storage Area Network

• Reservation of data for the system administrator
• Ten precepts of backup

Notes