Taobao dba organization's "iDataForum2010 China internet Database Technology Forum" Conference notes __ Database

A few days ago on micro-blog saw this "iDataForum2010 China Internet Database Technology Forum", signed up for. It's been a day today.

This technical forum is organized by Taobao DBA. Organizers hope to become a technical forum unrelated to specific enterprises. According to organizers, this is the second session, the first is held in Hangzhou.

Of course, the main DBA is to attend the meeting. Sat full of a chamber. There seems to be 200来 people. But after all, in Beijing is the first, and relatively professional, not very popular, not like Microsoft's TechEd, can sell books, training and other related industries units are called to stall.

I am interested in two of them to Taobao's database architecture as the theme of the speech, below I listen to some of the impressions, not necessarily accurate:

Taobao's database architecture has undergone a mysql->oracle->mysql process.

The original MySQL is a stand-alone version, relatively simple, the entire system using lamp.

After the business grew, the stand-alone version of MySQL became a bottleneck, and the system was upgraded to Oracle + IBM minicomputer + EMC high-end storage.

After the business growth, Oracle has become a bottleneck, Taobao first of the database to do a vertical split, divided into users, transactions, commodities and other databases.
The Oracle data is then replicated to a group of MySQL, where Oracle writes and MySQL reads. These MySQL have done the horizontal split again. This split is different from the two dimensions of the split, a group of MySQL is based on the seller to split the data, the other group of MySQL is based on the buyer to split the data. This way, whether by the buyer to check the data or according to the seller to check the data, will be relatively fast.

After the meeting organizers said the presenter's PPT will be put on the General Assembly website as soon as possible: http://www.idata-forum.org/