Source: chinaitlab Collection 2004-7-21 11:52:00
First, Oracle Data Guard
Oracle9i launched a work Guard. Can be powerful, more effectively implementing disaster recovery solutions Oracle Data
Oracle Data Guard By the library, the backup of the alternate database, the alternate database is immediately put into normal operation using the primary database, one or more management and monitoring work are automatically completed, so that the application is run without intermittent database backup The way the way is protected, when the primary database is down, there is at least to avoid the system's paralysis.
(1) Introduction to Data Guard
1. Database switch
Allows DBA to cut the primary database and the original database change to the user's request during the standby data. Convert to the alternate database, this alternate database library. This feature of Data Guard guarantees that the data will not be lost in response to the user's request, and the data is not lost, avoiding database recovery.
2. Enhance database availability by distributed configuration
Oracle Data Guard is called a site, usually natural disasters such as disasters, floods, and the database data will be made up to multiple spare databases, and the data in Data Guard is distributed around the Data Guard. With a network connection, even if you encounter an earthquake, the fire is well protected. The structure of the Data Guard is shown in the figure below:
3. Synchronize the data of the primary site and the standby site
In the Data Guard environment, a modification of a site is made to the form of the log transmission site with the primary site by the form of the archive log. Set to the primary site, used to respond to the user's request, the transaction automatically transmits the database from the primary site to each alternate site to achieve standby
4. Physical damage to the database
Since the physical damage to the primary site brings the risk of the database. It is impossible to transfer to the standby site by archiving log files, so it is reduced by physical damage.
(2) Switching in the database
Switch the primary database to the standby database, this counter. The database handover can be cut from the primary database role to the log role. Use the database to change to the primary database, and the original database is changed to the alternate database role, or switch from the alternate database role to
1. The working mode of the main database:
Guaranteed Protection:
It is stipulated that the disconnection of data is disabled when modifying the primary database, at least one alternate database is valid. And guarantee that there are no data loss. This mode is in the case of the connection between the active and standby, and the impact of the database performance is the most.
Instant Protection:
It is specified that the connection data between the active and standby is lost when modifying the primary database, and at least one alternate database is valid. Interrupt, allow data differences, and when recovery can have a small impact. Solve the phenomenon of data differences after connecting to the Guaranteed Protection mode. no
Rapid protection:
It is pointed out that the modification of the primary database is valid on the standby database. There is data loss and minimize the impact of database performance.
Delayed Protection:
It is pointed out that the Most Protection mode of the primary database is valid even if the loss of the network data is equivalent to the main data end alternate database. When the RAPID connection is valid, the primary database library is also allowed to do the number of uncountable logs. The most Protection and Delayed and all the spare databases have data differences, minimize the influence of database performance.
The difference between the four modes is shown in the following table:
2. Operating mode of the alternate database:
Managed Recovery Mode:
Maximize protection data, the primary database will be restored by the online server. Make log archive to the standby database, the alternate database automatically applies these read-only modes:
The standby database cannot apply the archive log. In this library, the main database continues to inquire the alternate database in the MOUNT mode. When the alternate data log archives on the standby database.
Although the alternate database cannot be in two modes in the GUARD environment, it can be restored in the restore management mode in two modes. Switch between schedules. In most DATA
3. The difference between Failover and Switchover
Failover:
Causes the main database OFFLINE, the alternate database online, which is caused by the system and software failure.
The Guaranteed Protection mode is applied even in the standby database. There may also be data loss, unless the alternate database is running
The instance must be restarted when the original primary database is reused.
Other alternate databases also need to restart instances.
Switchover:
Intent to synchronize the primary database OFFLIN. Such as: You can use SWIE, and another alternate database onlineetchover complete the smoothing upgrade of the system. , It can switch to the standby database without
Even if the redo log is not applied on the spare database, it will not cause loss of data.
The database does not need to restart realm to perform regular maintenance without example. This allows the primary database to be almost interrupt immediately. Restoring its function on the standby database, so
The difference between Failover and Switchover is: When the ability to lose the alternate database is lost, the database can be switched to the primary database, or from the master data failover, the alternate database is switched to the primary database, it can no longer return to the alternate mode And SwitchOver, the spare library switches back to the backup database.
4. Switch between primary database and alternate database
When the primary database operation is in the process of the GUA certificate database during the switch. If the archive log is not based, the database will cause the application of the database archive log to determine the data of the Ranteed Protection and INSTAN, which means the full application of the master database, or the main database works in R. The total amount of data loss. Can be determined. Tig PROTECTION Mode, you can protect all archived logs must be applied to the alternate number APID and DELAYED Protection Moderator Database Archive Log path properties and standby
Second, Oracle's Advanced Replication Technology
1. basic concepts
Copying, as the name refers to copy the number of databases in the database, which is copied for an over-distributed computing solution to a database of different physical locations to support distributed applications.
2. Basic structure of advanced replication technology
The physicalized view is in the previous OR master site. The ACLE version is called "snapshot". It is used to copy data to the replication environment
The physicalized view can be read-only, can be updated or writable.
(1) Read only physicalized view
In an infrastructure, a real site or a primary solidified view can directly send a read-only access table site directly to a read-only physicalized view. Applications can avoid accessing the maintenance data. The figure below shows read-only physical dependent data, which is derived from a main body site and does not consider whether the network is available, it
(2) Update physicalized view
In a more advanced structure, you can create a new entry, update, and delete rows on a row entry view on a new entity view. The following figure shows the use of updatable physicalized views that allow users to perform the same insertion, update, and delete primary table or main physicalized view by performing the same insertion, update, and delete a primary table or primary solidified view in this more operation.
3. Implement multiple subject copy selection
Synchronous copying, replication data is updated in the node's replication data that has been updated on any time for real-time requirements, and this change will be reflected in its commercial applications. Consistent. If you copy any of his replication nodes in your environment. This technique is suitable for asynchronous replication, and all replicated nodes have updated operations in the copy data of a node, which is changed. These different transactions can be spaced apart from a few seconds, and several data is temporary, but the propagation will eventually be safer during the fixed time. If one of the copied environments will be propagated and applied in different transactions into all other replication sections, hours, it can be a few days later. The data between all replicated nodes between the replication nodes is consistent.
Process replication, batch processing applications can be loaded on the network to the network, in order to avoid using process replication, it only changes a large amount of data in a single copy stores a single transaction. A typical row level is quenched, a process call to the batch application operation in the replication environment to accumulate data replicates.
Third, Oracle9i Database: Application Cluster Technology
Oracle9i is on the Internet in Oracle9i True Applications, Oracle9i really applies cluster bases on all computers on clusters, so that Oracle9i really applied Cache Fusion architecture to provide exciting reliability independent computers to deal with special Smoothability, performance, and reliability do not need to be redistributed, nor modifications to use these new increasing online application markets (Oracle9i Real Application at Oracle Get patented cache melting those frequently accessed data) To provide a cluster to provide a direct independent processing of more than four nodes, ORAC. Calculation tasks of the cluster technology provided by other vendors, the specificity of management data, and the specificity of the management data. So, in need to re-need Write an application, Oracle resource. Many critical improvements, its most special technology). As a new generation of cluster technologies as Oracle, it can quickly and effectively apply scalability. This breakthrough linear retractable Sex. On the other hand, with the LE9I true application cluster can be an e-commerce ratio, Oracle9i really applied clusters to use segmentation. " This cluster technology enables the system user cluster system to increase the computer from 9i real application clusters to transparent ways
1. Real Application Clusters architecture
Real application. Figure: STERS is composed of multiple components that can simultaneously access a shared database at the same time by multiple nodes
Real Application Clusters consists of the following components:
Cluster Manager
The Global Cache Service and Global Enqueue Service
Cluster Interconnect and Interprocess Communication (node-to-node)
Disk subsystems
In the Real Application CREAL Application Cluster to decompose into multiple small transactions, in different LUSTERS environments, all nodes can s ensure that each node accesses the node execution of shared data. It is suitable for DSS, OLTP, performs transactions, consistency, and integrity on the same database. Big transactions and hybrid systems can be used.
2. Oracle9iRealapplicationClusters features:
"Open box", near linear transparent zoom
Good compatibility with other procedures, no need to redesign
Rapidly growing cluster, quickly add nodes and disks
3. Hardware composition
Using cluster database technology,. Hardware and database data capacity is pusically piles into a reactor, such as Oracle 9i database mining databases, simultaneously improve system reliability on multiple servers: If a server fails, it will not increase the load. This feature is also able to take hardware (in some cases, in some cases, it provides excellent to save hardware investment and ensuring that enterprises have a problem of informationization of enterprises that have plagued the company's information construction. The "Cluster Technology" on a server (REAL APPLICA is running without having to increase the application code or body data capacity, and the company can cause damage by increasing the system, because other services shorten the database access time, thus improvement of expensive The mainframe is possible, this is millions of dollars). Oracle database to open performance, and leads today's database technology information exists in a single database. Enterprise legal solution is to continue to add a database. Now Different Tion Clusters), it enables a single system to make any modifications, this feature is greatly plus small servers to expand; and in any server can be very easy to share part of the new application performance. It is made to save nearly 80% of the CPU demand to save nearly 80%, 66% of the UNIX open system. The cache is highlight:
Oracle9i REAL ApplicCache Fusion is the cluster database cache responded. When data is positive, the normal reading clusters of data is guaranteed to use the significant breakthrough in new CACH technology. In the cluster, when you are updated, Cache Fusion can be taken and updated. E Fusion (Cache Welding) technology, the request can be associated between the cache on any server of the cluster database.
If a query request passes a high-speed cluster, "the scalability of the cache set. Responding to a remote cache, the data block welding" process will occur automatically. The process will be transparent from one node to another node, which greatly improves the group.
Oracle9i REAL APPLIC, using it quickly increases the network festival to provide transparent application scalability, almost and existing place. Oracle9i Real Applicatio resolves the network application compatible with the server in a process, which supports the cluster function n Clusters as a dispute between all servers in the cluster.
Oracle9i sets a new set of standards that these new features include disaster prevention, system errors, prevent system shutdown, which causes network interruption to ensure high availability of the system. Speed recovery and transparent recovery of human error
