Shulou(Shulou.com)06/01 Report--

The operation principle of DataGuard is very simple: transfer log, application log. The following figure shows the basic architecture of DG

The log transfer service transfers the log data generated by the master library to the slave database.

The application service (Apply Service) validates the log data and updates the data files from the library.

The write process of the master database updates the data file and does not depend on the DataGuard schema.

When the network or slave library fails to recover, DG automatically retransmits log data that has been archived by the master database.

Log transfer Service Redo Transport Services

Redo Transport Services coordinates log transfer between master and slave libraries. When the master library LGWR writes the log, the Log Network Server (LNS) program reads the log data from Log buffer cache and transfers the data to the slave library through Oracle Net Services. Receive incoming logs from Remote File Server (RFS) on the library. RFS writes the received log to standby redo log file (SRL). In the multi-slave library environment, the master library enables an independent LNS process for each slave library to transfer log data.

Using the LNS process, DataGuard supports two log transfer modes: synchronous and asynchronous.

Synchronous log transfer Synchronous Redo Transport

Synchronous transport (SYNC) is also known as "zero data loss". The commit operation on the master library is considered successful only when the LNS acknowledges that the log has been transferred to the slave library and has been written to disk.

Data Guard Log transfer process Architecture

SYNC redo transport architecture

The user commit a transaction and some redo record will be generated in the Redo Buffer. LGWR reads redo record from Log buffer and writes it to Online Redo Logs while waiting for confirmation from the LNS process.

LNS reads the same redo record from Redo Buffer and transfers it to the slave library through ONS. RFS receives the log and writes it to Standby Redo Logs.

When RFS writes the log successfully, it transmits a confirmation signal to LNS,LNS on the main library to inform LGWR of this confirmation information. At this point, LGWR can give feedback to the user about the successful submission.

Asynchronous transport (ASYNC)

The log is transferred asynchronously, and LGWR does not need to wait for the confirmation message of LNS. In this case, standby has almost no performance impact on the main library. If LNS cannot transfer logs faster than LGWR writes, when Log buffer is reclaimed, LNS is transferred from online redo file to the slave library from the read log. Once the speed of the LNS keeps pace, the data will be read from the Log buffer.

In one case, assuming that the number of the current log that LNS is reading is 10, two log switches occur during the re-read. Now the current log number is 12. By this time LNS had finished reading 10. It will not continue to read the No. 11 log. Instead, switch directly to the current log, numbered 12. In this case, the No. 11 log was not transferred to the standby database by LNS. We call it log file gap.

When log file gap occurs, the archived log files are transferred to the slave database by the arch process.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.