Oracle exadata Introduction

 

With the development of enterprise business, there are more and more large data warehouses, and their scale is also expanding rapidly, increasing by three times every two years on average. Large Data Warehouses require scanning dozens, hundreds, or thousands of disks at the highest disk read speed, this requirement can be met only when the pipeline bandwidth between the disk and the server is increased by 10 times or more. Therefore, enterprises often find that the larger the data warehouse, the slower the operation speed.

How can we break through the data bandwidth bottleneck? Oracle exadata, a new architecture, came into being. Oracle CEO Larry Ellison and Sun Microsystems Executive Vice President John Fowler announced in September 2009 that the world's first OLTP database machine-sun Oracle Database machine (Oracle exadata version 2) was launched ).

"The exadata database machine will be the most successful new product in Oracle's 30 years of development," Larry Ellison once gave high hopes for this product. The second version of Oracle exadata is a combination of Sun hardware and Oracle database software, while the first version is jointly created by Oracle and HP.

Sun Oracle Database machine adopts industry-standard hardware components, flash fire technology of sun, and 11g second edition of Oracle Database (Oracle Database 11g Release 2) and Oracle exadata storage server software version 11.2 (Oracle exadata storage server software release 11.2), when used for data warehouses, its running speed is twice that of the first version.

The bottleneck of the traditional data warehouse is that a large amount of data is read from the storage to the server memory. There are three methods to break through this bottleneck:
1. Reduce the amount of data transmitted
2. Added transmission pipelines.
3. Increase the width of the MPs queue.

 

Oracle exadata has made corresponding improvements to the above three aspects:

First, exadata itself has an independent CPU and memory, as well as an operating system (Oracle Enterprise Linux), which is actually a server. It provides an important feature: Smart scan. Smart scan is used by the database server to send query commands to exadata. exadata first filters the data in the storage, and then transmits the filtered results to the memory of the database server, this is very different from the traditional method. Transferring query processing to the storage system can greatly reduce the amount of data sent to the server.

Second, exadata is modular, and each exadata module is called a storage unit. As the number of storage units increases, the storage throughput of the entire cabinet increases. You can even connect multiple cabinets to provide higher throughput. It is a large-capacity parallel storage grid. By adding "storage units", you can increase the number of transmission pipelines.

Again, the transmission interface of exadata is InfiniBand, an oracle-proud interconnection technology standard, 10 times faster than traditional fiber-optic channels. This is like 10 times the width of the transmission pipeline.

Oracle exadata version 2 provides four types of configuration: entire server rack (8 database servers and 14 storage servers), half-Rack (4 database servers and 7 storage servers) 1/4 racks (2 database servers and 3 storage servers) and Basic Systems (1 database server and 1 storage server ). The above four exadata configurations can be available immediately. With only the wires added, the sun Oracle Database machine can be expanded to eight racks, meeting the needs of various application scenarios.

 

Official website address:

Http://www.oracle.com/cn/products/database/exadata/index.html

 

 

 

A Daniel's article is detailed:

Analysis of Oracle exadata Technology

Http://www.hellodba.net/2010/02/oracle_exadata.html

 

 

 

Organized from Network

------------------------------------------------------------------------------

Blog: http://blog.csdn.net/tianlesoftware

Online Resources: http://tianlesoftware.download.csdn.net

Video: http://blog.csdn.net/tianlesoftware/archive/2009/11/27/4886500.aspx

Dba1 group: 62697716 (full); dba2 group: 62697977 (full)

Dba3 group: 62697850 super DBA group: 63306533;

Chat group: 40132017

-- Add the group to describe the relationship between Oracle tablespace and data files in the remarks section. Otherwise, the application is rejected.