|
Dual-machine Hot Standby topic: Introduction |
With the continuous advancement of information construction, the activities of various enterprises and institutions are increasingly dependent on their key business information systems. These business information systems play a vital role in the operation and development of the entire organization, in the event of downtime failure or application downtime, it will bring huge economic losses to the Organization.
It can be seen that for those organizations that need to ensure information security and provide uninterrupted information services, Fault Tolerance and Non-disconnection of business systems are particularly important. How to ensure the continuous operation of various key applications and achieve a virtuous circle of sustainable operation has become a key issue that enterprises and it institutions urgently need to solve. |
|
Definition of Dual-machine Hot Standby: |
Hot Standby includes two types: generalized and narrow.
In a broad sense, it is another term for high-availability server applications. The English translation is "High Available". What we usually call Hot Standby is translated and belongs to the high-availability category, the dual-host hot standby only limits two highly available servers. Hot Standby software is a software solution for solving an inevitable plan and unplanned system downtime problem. Of course, it also has hardware. Is the Basic software for building highly available cluster systems. For any faults that cause system downtime or service interruption, the software process is triggered for error determination, fault isolation, and local online recovery to continue executing the interrupted service. In this process, the user only needs to experience a certain degree of acceptable latency, and can restore the service in the shortest time.
In a narrow sense, dual-host Hot Standby refers to hot standby (or high-availability) based on two servers in a high-availability system. Because the two servers are widely used in China, therefore, the name is "Hot Standby". The active/standby mode is divided into active/standby mode and active/standby mode ), the master-slave mode means that a server is inA serviceIs active, and the other server is in standby (standby) of the service ). The dual-Host Mode refers to the active-standby and standby-active statuses of two different services on the two servers respectively ).
Note: the active-standby status refers to the status of an application or service, not the server status. |
|
There are two main methods to form dual-machine Hot Standby: |
1. shared storage (disk array:
Shared storage ensures data integrity and continuity after switching through disk arrays. User Data is usually stored on the disk array. When the host is down, the slave machine continues to obtain the original data from the disk array. As shown in.
This method is often called a single disk failure by industry insiders because of the use of a storage device. However, the storage is generally highly secure. Therefore, if you ignore the storage device failure, this method is also the most popular hot backup method in the industry.
2. Data Replication:
This method mainly uses data synchronization to ensure data consistency between the master and slave servers. |
|
There are two main methods to form dual-machine Hot Standby: |
1. shared storage (disk array:
Shared storage ensures data integrity and continuity after switching through disk arrays. User Data is usually stored on the disk array. When the host is down, the slave machine continues to obtain the original data from the disk array. As shown in.
This method is often called a single disk failure by industry insiders because of the use of a storage device. However, the storage is generally highly secure. Therefore, if you ignore the storage device failure, this method is also the most popular hot backup method in the industry.
2. Data Replication:
This method mainly uses data synchronization to ensure data consistency between the master and slave servers.
There are several methods for data replication, and their performance and security are also different. The main methods are as follows:
A. The document-based test bay is not applicable to applications such as databases, because opened files cannot be copied. If you want to copy files, you must disable the database. This is obviously not acceptable. File Replication is mainly applicable to web page updates, FTP uploading applications, data integrity on the master and slave servers, and use cases where the continuity requirements are not high.
B. Take advantage of the database's copy function, such as the subscription replication feature of sqlserver2000 or 2005. This method should be carefully used by users based on their own applications, mainly because:
(1)'s review copy adds fields to the user table, which has high requirements on application programming. If fields are not explicitly specified during application software writing, however, using this function will make the application unable to work normally.
(2) Data Retention: this restriction is also the most terrible, because sqlserver does not arrive at the master and backup machines in real time during data transmission, but writes data to the host first, write the data to the backup machine. As a result, the data of the backup machine is often too late to be updated. In this case, if a switchover occurs, the data of the backup machine will be incomplete and discontinuous, if the user finds that the written data cannot be found on the slave machine and re-writes the data, after the host is repaired, there will be a serious conflict of data between the master and slave machines, and the database will crash.
(3) For complex applications, do not use subscription replication for dual-host hot backup, including processing of stored procedures, triggers and sequences in the data structure. In the event of a conflict, modification is very troublesome.
(4) The server performance is reduced. For larger databases, the subscription replication of sqlserver2000 or 2005 will slow the server database operation.
In short, the subscription replication of sqlserver2000 or 2005 is mainly used in the data snapshot service, and it is not used for data synchronization in hot standby.
C: Hard Disk Data Interception. At present, mature dual-host Hot Backup Software in China and abroad usually uses the hard disk data interception technology, which is usually called the mirror software or the mirror software, this technology is now very mature, and the interception methods are also different.
(1) Partition interception technology. Taking the pluswell Hot Backup product as an example, it adopts a partition hard disk sector interception technology through driver-level interception, extract the data written to the hard disk and write it to the backup server first to ensure that the data of the backup server is up-to-date, and then write the data back to the host hard disk. This method ensures that the data in the master-slave database is completely consistent, and the database integrity and continuity can be ensured regardless of the switchover. Because partition interception technology is used, users can divide suitable partitions on a hard disk as needed to complete data synchronization.
(2) Hard Disk interception technology. Taking Symantec's co-standby as an example, it is also an effective hard disk interception software. Its interception is mainly based on a whole hard disk, hard Disk initialization usually takes a lot of time. |
|
Articles on Dual-machine Hot Standby Solutions |
Introduction to dual-machine hot standby and dual-machine hot standby system solution for a power plant
· High Availability Dual-host LAN Hot Standby Solution
· Hot Standby solution for large and medium-sized general hospitals
· Shenzhou digital firewall dual-host Hot Standby Solution
· Dual-machine Hot Standby solution for Shanxi Mobile text message Platform
· Typical storage backup Topology |
|
Conclusion |
| The advantage of using a dual-machine hot standby system is that for important services, two servers are used to back up each other and execute the same service together. When one server fails, the other server can assume service tasks, so that the system can automatically guarantee continuous service without manual intervention. The standby server solves the problem of uninterrupted service when the primary server fails. It is server-level fault tolerance. |
|
