Source: daily security net
An image is an information storage process that generates an image view of the same data on two or more disks or disk subsystems. One is the primary image system, and the other is the backup image system. The master-slave image storage system can be divided into local images and remote images. The master-slave Image Storage System of the local image is in the same RAID array, while the master-slave Image Storage System of the remote image is usually distributed across different nodes of the man or WAN.
Remote images, also known as remote replication, are the core technology of disaster tolerance backup and the basis for maintaining remote data synchronization and disaster recovery. It uses the remote data connection function of storage devices separated from physical locations to remotely maintain a set of data images. In the event of a disaster, data backups distributed in remote storage are not affected. Whether the host requesting the image requires confirmation from the remote image site can be divided into synchronous remote image and asynchronous remote image.
When the core process of an enterprise fails to work properly due to a disaster, it will cause some losses to the enterprise. Such loss may be quantifiable, such as the loss of documents, the direct loss caused by calculation errors, or intangible loss, such as the loss of customer satisfaction and competitive advantages. Based on the comprehensive consideration of quantifiable and non-quantifiable losses, the decision-making basis for various core business processes due to disaster damage tolerance and loss is obtained. It is reflected in the IT system and measured mainly through the following three indicators: ① data recovery point objective (RPO): shows the tolerable degree of data loss when the process recovers from a disaster; ② recovery time objective (RTO): indicates the urgency of the process after a disaster occurs (that is, how long it will take to recover); ③ network recovery objective (NRO ): that is, when the business outlets can resume communication between the backup network and the data center.
For RPO and RTO alone, the score for Synchronizing Remote images is undoubtedly the highest. Synchronous remote image (synchronous replication technology) refers to the use of remote image software to copy local data to a remote location in full synchronization mode, each local I/O transaction must wait for the confirmation information of the remote replication to be completed and released. Synchronous images make the remote copy always match the content that the local machine requires to copy. When the primary site fails, your applications are switched to the backup site, and the remote copy of the image ensures that the Service continues to run without data loss. In other words, the RPO value for Synchronizing Remote images is zero (that is, no data is lost), and RTO is also measured in seconds or units. However, since round-trip transmission may cause a long delay and the performance of the local system is directly linked to the remote backup device, synchronous remote images are only applicable to applications that are relatively close to each other, generally, the interval between master and slave Image Systems cannot exceed 160 kilometers (about 100 miles ).
Asynchronous remote image (asynchronous replication technology) is provided by the local storage system to request the I/O operation of the image host to complete the confirmation information, ensure that the basic operations for outputting/inputting data to the local storage system are completed before updating the remote storage view, that is, its RPO value may be calculated in seconds, it may also be measured in minutes or hours. It adopts the "store-and-forward" technology, and all I/O operations are performed synchronously in the background, which makes the local system performance Suffer little, this greatly shortens the waiting time for data processing. Asynchronous remote images have the advantages of "low network bandwidth requirements and long transmission distance (up to 1000 kilometers. However, many remote "write" operations from the image system are not confirmed. When data transmission fails for some reason, it is very likely that the data consistency between the master and slave systems will be damaged.
The biggest advantage of synchronous remote images and asynchronous remote images is to reduce the risk of data loss caused by disasters to a minimum (asynchronous) or even zero (synchronous) level. Second, in the event of a disaster, it takes a short time to recover a process. This is because you do not need to use a proxy server to create a remote data image. It supports heterogeneous servers and applications.
The price of remote image software and related equipment is generally high, and it must occupy at least two times of the primary disk space. However, if your company's business process itself requires a high level of data loss (RPO) or recovery time (RTO), it is the best solution to establish a remote image.
In addition to being expensive, remote image technology also has a fatal defect that cannot prevent disasters such as system failure (rolling disaster), data loss, damage, and accidental deletion. If the data on the master site is lost, damaged, or deleted by mistake, the data on the backup site will also have a chain reaction. Currently, only a few asynchronous remote image products on the market can stamp timestamp for each transaction. In the event of data corruption or accidental deletion, you can specify the status of data recovery to a certain time point, of course, it is not enough to install the remote image software to implement this function. You also need to take other necessary protection measures, for example, the latency replication technology (local data replication is performed in the background log area), and remote data update is performed after local data is intact. In addition, remote image technology also has many disadvantages, such as the inability to support heterogeneous disk arrays and internal storage components, the lack of supported software types, and the inability to provide file information.
If you have high requirements on the RPO and RTP metrics and are unable to afford the remote image solution, is there any other feasible alternative? Of course, there are products based on technologies such as continuous data protection (CDP), time-based continuous snapshot replication, automatic backup, automatic data replication and update, and distributed backup. They support heterogeneous storage systems and rollback (roll back the current transaction and cancel all updates in the current transaction), but their total cost of ownership (TCO) is lower than that of remote image products, in addition, the "Install and manage proxy servers" step is added.