Comments on: Data Warehousing with MySQL

By: Okky Hendriansyah Tri Firgantoro

Okky Hendriansyah Tri Firgantoro — Tue, 25 Jul 2006 09:03:42 +0000

Hi Nishith, I really enjoy readingyour posts.
I am interested with the technical detail of the concept. Where can I look for it?

By: Nishith

Nishith — Wed, 12 Apr 2006 03:21:53 +0000

Hi Yishan,
Not really. Since Data Warehouses do not have constant insert/updates ( as required for OLTP systems), MyISAM tables do quite well and shouldn’t get corrupted unless the ETL job fails in some very weird way (server restart, hardware crash, etc.).

Even then, the corrupted portion would be right at the end (Data Warehouse ETL is usually an append process) which makes the repair simpler (you can just throw out all the data from the last load job).

Of course, it’s a good habit to keep checking the table periodically (CHECK TABLE or MYISAMCHK).

By: Yishan

Yishan — Tue, 11 Apr 2006 20:29:22 +0000

Aren’t MyISAM tables more prone to corruption though? InnoDB can do a considerable amount of self-repair, which is useful when the largest (and usually the most valuable) tables crash, while MyISAM would then require you to write a supporting backup/redundancy system to handle crashes.

By: Nishith

Nishith — Thu, 03 Nov 2005 21:11:53 +0000

Thanks RT. I am sure we’ll be able to do away with them silly ceiling fans.

By: RT

RT — Thu, 03 Nov 2005 14:01:08 +0000

Look forward to the comprehensive analytics project, I think it is something that everyone who is into data warehousing will be interested in. I would love to see my reports fly, without help from ceiling fans