发信人: NNYYCC (牙膏皮换糖), 信区: Database
标 题: Re: Re: 问个database design的问题
发信站: BBS 未名空间站 (Mon Jun 18 11:16:09 2007)
I believe table partition will be a good option for your data. Also, create
an index on stock_id AFTER table is partitioned.1000 stocks for a multi-
million row table is still considered low cardinality and an index will
improve the performance.
【 在 Fine (coulda shoulda woulda) 的大作中提到: 】
: Really appreciate it!
: OK. back to the question.
: Here, assume I'm doing research on stock price history. I have tick data
: 1000 stocks, each with 1 million ticks, in separate text(.csv) files, one
: file per stock.
: I would like to import the data (csv files) into database and run queries
: the DB, and I only run query on one stock at a time. This is not really a
: transactional database, if I understand what it is -- the import part
: happens very infrequently, while I need to do a lot of queries on each
※ 来源:·WWW 未名空间站 海外: mitbbs.com 中国: mitbbs.cn·[FROM: 156.145.]