[Hadoop in China 2011] HBase applications and improvements on the Taobao platform experience

http://tech.it168.com/a2011/1202/1283/000001283152.shtml

In the NoSQL systems and applications sub-forum this afternoon, a software engineer from Taobao Chang keynote "Taobao HBase applications and improvement", describes the HBase use on the Taobao platform, as well as HBase debugging and improved experience sharing .

HBase applications and improve the experience on Taobao platform
▲ Taobao software engineer Zhang Yi

   With the rapid growth of Taobao data, the amount of data to double every year, this growth rate is Taobao selected HBase important reason. According to Zhang Yi introduction, Taobao since the beginning of this year to HBase rehearsal, including online applications and tuning, and bug changes. The process includes three stages, the first stage of the process is familiar with other performance attributes and configuration HBase; a second stage HBase applied to the line; the third stage is to modify the codes, and communities into HBase and allowed to be further developed.

HBase applications and improve the experience on Taobao platform
The reason HBase ▲ Taobao selection

   NoSQL HBase distributed storage system, which is based on the underlying data store the HDFS, having a high availability and performance characteristics, can process ten billion one million rows * version * thousands of storage columns. About why Taobao will choose HBase, Chang said, Taobao selected as Hadoop HBase just as she chose, because of its ability to mass data processing. Taobao amount of data each day to achieve the PB level, need HBase challenges brought about huge amounts of data. Further, HBase has an easy to scale, strong consistency characteristics of random access performance, high reliability and stability, single write.

HBase applications and improve the experience on Taobao platform
▲ HBase scale Taobao

   Chang believes that a large amount of data written to the moment, the database need not support or support of a high cost scenario may consider using HBase, or need long-term preservation of data, and the amount will continue to grow to larger scenes can be selected HBase. But after all, not a relational database HBase, and therefore does not apply to complex Join, multi-level indexes, table relational data model.

HBase applications and improve the experience on Taobao platform
▲ NoSQL special tribute to sitting on the ground full of students

Reproduced in: https: //www.cnblogs.com/licheng/archive/2011/12/05/2276414.html

Guess you like

Origin blog.csdn.net/weixin_33724570/article/details/92626699