Data backup and data disaster-tolerance analysis

data backup and data disaster-tolerance analysisData disaster recovery and data backup the key data loss will disrupt the normal business operation of enterprises, resulting in huge economic losses. To protect data, businesses need to back up disaster-tolerant systems. But a lot of companies in the backup system after the think of peace of the matter, in fact, need to build disaster recovery system. The relationship between data disaster recovery and data backup is mainly embodied in the following aspects: 1. Data backup is the base of data disaster recovery is the last line of defense for data high availability, the purpose of which is to recover data quickly when the system data crashes. Although it is also considered a disaster-tolerant scheme, the disaster-tolerant capacity is very limited, because the traditional backup is mainly the use of data built-in or external tape drive for cold backup, backup tapes are also in the computer room unified management, once the entire room has been a disaster, such as fire, theft and earthquakes and other disasters, these backup tapes will also be destroyed,      The stored tape backup also does not have any disaster recovery capabilities. 2. Disaster tolerance is not a simple backup real data disaster is to avoid traditional cold backup is congenitally deficient, it can in the event of disaster, full and timely restoration of the entire system. Disaster tolerance can be divided into several levels according to its capacity of disaster tolerance, such as the international standard share 78 disaster tolerance system defined by seven levels: from the simplest only local tape backup, to the backup of the tape stored offsite, and to establish the application system real-time switching off-site backup system, Recovery times can also be from days to hours to minutes, seconds, or 0 of data loss.
Regardless of the disaster recovery scheme, data backup is the most basic, there is no backup data, any disaster-tolerant scheme has no practical significance. But it is not enough to have a backup, and disaster tolerance is also essential. Disaster tolerance for it is to provide a computer information system to prevent all kinds of disasters. Technically, there are two main indicators for measuring disaster tolerance systems: RPO (Recovery point object) and RTO (Recovery time object), where RPO represents the amount of data that is allowed to be lost when a disaster occurs, and RTO represents the timing of system recovery.      3. Disaster tolerance is not only a technology     disaster tolerance is a project, not just technology. At present, many customers still stay in the attention of disaster-tolerant technology, and the disaster-tolerant process, norms and specific measures are not clear. It also never evaluates the feasibility of disaster-tolerant schemes, and believes that it is very risky to have a disaster-tolerant plan in place. Especially in some small and medium-sized enterprises, think their enterprises for data backup and disaster, the whole year spent a lot of human and financial resources, and the results of a few years there is no major disaster, so relaxed vigilance. But in the event of a disaster, regret is too late. This foreign multinational company is doing very well, despite the fact that there have been no major disasters in the past few years, backup of the tape, almost no use, but still, very serious to do every step, and basically every month on the feasibility of the current disaster recovery program assessment, field exercises.     II, data disaster tolerance level     design a disaster recovery backup system, you need to consider a number of factors, such as backup/restore data volume, the application of data center and the data center and the distance between the database and data transmission mode, The rate of recovery required in the event of a disaster, the management of the reserve Centre, and the funds invested. According to these factors and different applications, the common disaster recovery level has the following four:      No. 0 Level: Local backup, local save cold backup     This level disaster recovery backup, is actually the data backup referred to above. Its disaster recovery capacity is the weakest, it only local data backup, and the backup of the data tapes are only stored locally, not sent offsite.    in this disaster-tolerant solution, the most commonly used device is the tape drive, of course, depending on the actual needs of the tape drive can be manually loaded, you can automatically load the tape drive. The former is mainly used for small and medium-sized enterprises with small storage data capacity, the main products are IBM's Totalstorage Ultrium external tape drive 3580, single box disk can reach 200GB, pressureAfter shrinking can store 400GB data, can meet the majority of small and medium-sized enterprises, and even large enterprises data backup needs. HP's tape drive product line is very comprehensive, and the manually loaded tape drive currently consists of HP SureStore dat 24, hp SureStore dat 40, HP StorageWorks DAT 72, the capacity of which the single tape can store is 24GB, 40GB and 72GB, different users have the flexibility to choose, in addition HP also has a lto-ultrium tape drive series tape drive and DLT/SDLT series tape drive.     automatically load the tape drive IBM mainly has totalstorage Enterprise tape drive 3592,HP has StorageWorks DAT 40x6t and so on.      Note: The above is only the IBM and HP two well-known brands of products as a representative to explain, in fact, there are many other brands of tape drive products can be selected, such as Hitachi, Sony, Fujitsu and so on. In addition to selecting a tape drive, you can also select a storage device, such as a tape library, a jukebox, and a CD-ROM, for local backup storage.     level 1th: Local backup, offsite cold backup     backup critical data locally, then send it offsite for safekeeping. After a disaster occurs, restore the system and data according to the scheduled data recovery program. This disaster-tolerant program is also used for local backup with storage devices such as tape drives, which are selected for the previous introduction. You can also select storage devices such as tape libraries, optical libraries, and so on.     we often see companies in order to avoid tape dividend stolen and destroyed by backup tapes due to room security problems, it is often necessary to put backup tapes, especially more than a month's backup tapes, into specialized safes, and even to rent a bank's specialized safe to store these backup tapes. But this cannot be said to be a loss, because, in general, the location of the storage tapes and the company in the same city, in case of earthquakes, war and other natural disasters, these backup disk is still difficult to escape the bad luck.      level 2nd: Hot backup Site Backup      set up a hot backup point in the off-site, data backup over the network. That is, through the network in a synchronous or asynchronous way, the primary site data backup to the backup site. Backup sites generally only back up data, do not assume business, topological structure. When a disaster occurs, the backup site takes over the business of the primary site, thereby maintaining continuity of the business operation.     the disaster-tolerant location of remote data disaster-tolerant scheme in different places is usually chosen at a distance of not less than 20 km range, with the same configuration as the local disk array, through the optical fiber dual redundancy access to the San network, the implementation of the local key application data real-time synchronous replication. In the case of local data and a disaster for the entire application system, the system maintains at least one of the key business-critical data that is available offsite. This data is a fully real-time copy of the local production data. For the enterprise network, the data disaster recovery system is composed of the main data center and the backup data center.     The main data center is designed with a high reliability cluster solution, and the backup data center is connected to the primary data center through the fiber. The main data center system configuration host consists of two or more servers and other related servers, and the HA software is installed to form a high reliability environment of multiple machines. The data is stored in the primary data center storage disk array. At the same time, in the offsite Backup data Center configuration of the same structure of the storage disk array and one or more backup servers. With dedicated disaster recovery software, you can automate real-time, full backups of the primary data center's storage data and backup data center data. In the primary data center, you can also configure a tape backup server to install backup software and tape libraries, as required by the user. Backup servers are directly connected to storage arrays and tape libraries, which control tape backups of the system's day-to-day data. Among them, optical transmission equipment as the transmission equipment between two data centers, two data centers use optical transmission devices to compose optical self-healing ring through optical fiber, which can provide up to 80G (protection) and 160G (unprotected) communication bandwidth.      level 3rd: Active Mutual Assistance Backup     This off-site disaster recovery program is similar to the hot backup site backup scenario described earlier, except that the main, from the system is no longer fixed, but each other's backup system. The two data center systems are built at a distance, and they are both working and backing up data from each other. When a data center disaster occurs, another datacenter takes over its task. Usually in the two systems in the fiber-optic device connection also provides redundant channels, in case the work channel failure in time to replace the work, of course, this disaster-tolerant way is mainly the financial strength of large enterprises and telecommunications-grade enterprises.     this level of backup according to the actual requirements and the amount of money invested, but also divided into two:     ① two data centers are limited to the mutual backup of key data;
    ② Two data centers mirror each other, that is, 0 data loss. 0 data loss is the most demanding disaster-recovery backup, which requires that the system ensure data security regardless of any disaster. Therefore, it needs to configure complex management software and dedicated hardware equipment, the need for investment is relatively the largest, but the fastest recovery rate.     the two methods of hot backup are no longer traditional tape cold backup, but advanced channel technologies such as Sans, synchronizing server data, or asynchronous storage (mirroring) on remote dedicated storage devices (or tape devices).     in the above two kinds of thermal backup disaster recovery scheme, the main equipment includes disk array, fiber switch or disk machine. Is HP's low grade disk array hp StorageWorks MSA1000, a high-end virtual disk array hp StorageWorks VI 7410 product.     IBM's enterprise-class fibre switch--totalstorage®san switch F32 offering 32 fibre exchange ports; it's a mid-range fibre switch--totalstorage®san switch F16, It can provide 16 fibre exchange ports.     Iii. Disaster Tolerance technology     in the establishment of disaster recovery backup system will involve a variety of technologies, such as: San or NAS technology, remote mirroring technology, virtual storage, ip-based San Interconnection technology, snapshot technology.      1. Remote mirroring technology     remote mirroring technology is used in data backup between the primary data center and the standby center. A mirror is an information stored procedure that produces a mirrored view of the same data on two or more disk or disk subsystems, one called the primary mirror system and the other from the mirror system. The location in which the master-slave mirror storage System is located can be divided into local and remote mirrors.     remote mirroring, also known as remote replication, is the core technology of disaster recovery, and it is also the basis of keeping remote data synchronization and achieving disaster restoration. Whether the remote mirror-requested host requires a remote mirror site's acknowledgment information, and can be divided into synchronous remote mirrors and asynchronous remote mirrors.     Synchronous remote Mirroring (synchronous replication technology) refers to the remote mirroring software, the local data in a fully synchronized way to the offsite, each local I/O transactions need to wait for remote replication of the completion of the confirmation information before release。 Synchronous mirroring makes remote copies always match the content that the local machine requires to replicate. When the primary site fails, the user's application switches to the backup replacement site, and the remote copy that is mirrored can guarantee that the business continues without losing data. However, it has the disadvantage of long time delay, which is limited to the application in relatively close distance.     Asynchronous remote Mirroring (asynchronous replication technology) ensures that basic I/O to the local storage system is completed before the Remote Storage view is updated, and that the I/O operation by the local storage system to the requesting mirror host completes the acknowledgment information. Remote data replication is in the background synchronization, which makes the local system performance is very small, long transmission distance (up to 1000 km), the network bandwidth requirements small. However, many remote slave storage subsystem writes are not acknowledged, and data consistency issues may occur when a certain factor causes a transfer failure. In order to solve this problem, most of the technology is delayed replication, that is, to ensure that local data is intact after the remote data update.      2. Snapshot technology     Remote mirroring technology is often combined with snapshot technology to achieve remote backup, that is, by mirroring the data back to the Remote Storage system, and then using the snapshot technology to back up the Remote Storage System information to the remote tape library, CD-ROM library.     snapshots are a quick scan of the software for the disk subsystem to be backed up, a snapshot logical unit number LUN and snapshot cache to back up data, and a quick scan to quickly copy the blocks of data that are about to be modified in the backup process to the snapshot cache. A snapshot LUN is a set of pointers that point to the unchanged block of data in the snapshot cache and the disk subsystem (during the backup process). At the same time as normal business, use the snapshot LUN to achieve a full backup of the original data. It can enable users to extract the current online business data in real time without any impact on the normal business. Its "Backup Window" is close to 0, can greatly increase the continuity of the system business, to achieve a truly 7x24 operation of the system to provide a guarantee.     snapshot is a buffer (snapshot cache), which provides the real-time data image of system disk storage by the snapshot software, which has the problem of buffer scheduling.      3. Interconnect technology     Data backup between early primary and standby data centers, primarily based on San remote Replication (mirroring), which connects two sans via Fibre Channel FC for Remote Mirroring (replication). When a disaster occurs, the backup data center replaces the primary data center to ensure continuity of the system's work. This remote disaster preparednessThere are some defects in the method, such as: high cost, poor interoperability of equipment, short geographical distance (10 km), etc., these factors hinder its further popularization and application.     at present, a variety of IP-based SAN remote data disaster backup technology has emerged. They are remote replication of information from the primary Data center San through an existing TCP/IP network through an ip-based San interconnect protocol to a backup center SAN. When a backup center stores a large amount of data, it can be backed up to a tape library or a CD-ROM using snapshot technology. Remote disaster recovery backups of this ip-based San can span LAN, man, and WAN, with low cost and scalability, with broad prospects for development. Ip-based interconnection protocols include: FCIP, IFCP, InfiniBand, iSCSI, and so on.      4. Virtual storage     in some disaster-tolerant solutions, virtual storage technology is also adopted, such as Cyril Disaster Disaster recovery program. The virtualization storage technology has created a new situation in the system elasticity and scalability. It concatenates different storage devices, such as several IDE or SCSI drives, into one storage pool. The entire storage capacity of a storage cluster can be divided into multiple logical volumes and managed as a virtual partition. Storage thus becomes a function rather than a physical property, which is the primary limitation of the server-based storage structure.     Virtual Storage systems also provide the ability to dynamically change the size of logical volumes. In fact, the capacity of the storage volume can be increased or decreased freely online. You can change the size of a logical volume in a cluster by increasing or decreasing the number of physical disks in the system. This feature allows the volume's capacity to change dynamically with the user's immediate requirements. In addition, storage volumes can easily change capacity, move, and replace. When you install the system, you only need to allocate the minimum capacity for each logical volume and leave the remaining space on the disk. With the development of the business, you can use the remaining space to extend the logical volume as needed. You can also transfer data online from the old drive to the new drive without disrupting the operation of the service. A key advantage of     storage virtualization is that it allows heterogeneous systems and applications to share storage devices regardless of where they are located. The company will no longer need to connect one tape device to the server on each division.