Comments on: Dumb, and faster, databases http://labnotes.org/2008/08/23/dumb-and-faster-databases/ Sun, 15 Jan 2012 21:29:54 +0000 hourly 1 http://wordpress.org/?v=3.3.1 By: Assaf http://labnotes.org/2008/08/23/dumb-and-faster-databases/comment-page-1/#comment-141243 Assaf Wed, 27 Aug 2008 18:03:54 +0000 http://blog.labnotes.org/?p=1175#comment-141243 Generally, higher normal forms would be slower to query: there's more work involved because the data is spread over more rows, and query optimizers are only so smart. But higher normal forms is where the relational model shines. Like, Apotheon said, if you don't care about higher normal forms, perhaps you shouldn't be using an RDBMS to begin with. If the best way to manage your data is 3NF (or 4NF or 5NF), then you're willing to put up with less performance for the added benefits. What I see a lot of people doing, is using RDBMS because they need to store data. Fail. Without understanding what RDBMS are good for you end up misusing them. On one end we have cases where the RDBMS adds unnecessary complexity that slows you down (performance and development). On the other hand, poor schemas that don't take advantage of the relational model. Generally, higher normal forms would be slower to query: there’s more work involved because the data is spread over more rows, and query optimizers are only so smart. But higher normal forms is where the relational model shines.

Like, Apotheon said, if you don’t care about higher normal forms, perhaps you shouldn’t be using an RDBMS to begin with. If the best way to manage your data is 3NF (or 4NF or 5NF), then you’re willing to put up with less performance for the added benefits.

What I see a lot of people doing, is using RDBMS because they need to store data. Fail.

Without understanding what RDBMS are good for you end up misusing them. On one end we have cases where the RDBMS adds unnecessary complexity that slows you down (performance and development). On the other hand, poor schemas that don’t take advantage of the relational model.

]]> By: apotheon http://labnotes.org/2008/08/23/dumb-and-faster-databases/comment-page-1/#comment-141242 apotheon Wed, 27 Aug 2008 17:38:41 +0000 http://blog.labnotes.org/?p=1175#comment-141242 Unfortunately, Stu, if you're running a database in first normal form, you're running an almost random pile of data -- at least as far as maintenance is concerned. That's not exactly something I'd care to deal with over long periods, except in data sets so small I'd be better off using a flat file anyway (like Firefox's cookie policy exceptions database). Unfortunately, Stu, if you’re running a database in first normal form, you’re running an almost random pile of data — at least as far as maintenance is concerned. That’s not exactly something I’d care to deal with over long periods, except in data sets so small I’d be better off using a flat file anyway (like Firefox’s cookie policy exceptions database).

]]> By: Ole Phat Stu http://labnotes.org/2008/08/23/dumb-and-faster-databases/comment-page-1/#comment-141240 Ole Phat Stu Wed, 27 Aug 2008 14:38:55 +0000 http://blog.labnotes.org/?p=1175#comment-141240 Could it be that 3rd normal form is slower than 2nd is slower than 1st? And 3NF is what most people use/undertsand by a RDBMS? Could it be that 3rd normal form is slower than 2nd is slower than 1st?
And 3NF is what most people use/undertsand by a RDBMS?

]]> By: Chipping the web: August 26th -- Chip’s Quips http://labnotes.org/2008/08/23/dumb-and-faster-databases/comment-page-1/#comment-141238 Chipping the web: August 26th -- Chip’s Quips Wed, 27 Aug 2008 14:00:55 +0000 http://blog.labnotes.org/?p=1175#comment-141238 [...] Labnotes » Dumb, and faster, databasesPrecisely why Synergy DBMS still powers some of the heaviest used apps in the world.Tags: sql performance database [...] [...] Labnotes » Dumb, and faster, databasesPrecisely why Synergy DBMS still powers some of the heaviest used apps in the world.Tags: sql performance database [...]

]]> By: Jason Watkins http://labnotes.org/2008/08/23/dumb-and-faster-databases/comment-page-1/#comment-141236 Jason Watkins Wed, 27 Aug 2008 09:41:18 +0000 http://blog.labnotes.org/?p=1175#comment-141236 Assaf: I agree with the general idea, however MySQL's query planner is particularly dumb, and it's lack of certain features taken as given for other database engines (ie hash join) amplifies the problem. Assaf: I agree with the general idea, however MySQL’s query planner is particularly dumb, and it’s lack of certain features taken as given for other database engines (ie hash join) amplifies the problem.

]]> By: Assaf http://labnotes.org/2008/08/23/dumb-and-faster-databases/comment-page-1/#comment-141224 Assaf Tue, 26 Aug 2008 00:12:55 +0000 http://blog.labnotes.org/?p=1175#comment-141224 "Jeremy Zawodny is an incoming employee of Craigslist, having left Yahoo!'s platform engineering group. He has been described as 'Yahoo!'s MySQL guru'" I'm just saying (source: Wikipedia, so you know it must be true!) The problem is not RDBMS, but that people hold on to certain "truths" that are wrong. Computers are very lousy in playing along to our believes. If you know how the query optimizer works and the algorithms it uses, you know why (and when and where) it's wrong. I think most people will find it easier to relate to a real life example, rather than discussing the internals of query optimizers, which is why I chose to link to this post. And the problem is not isolated to MySQL, all the major database servers have these issues. “Jeremy Zawodny is an incoming employee of Craigslist, having left Yahoo!’s platform engineering group. He has been described as ‘Yahoo!’s MySQL guru’”

I’m just saying (source: Wikipedia, so you know it must be true!)

The problem is not RDBMS, but that people hold on to certain “truths” that are wrong. Computers are very lousy in playing along to our believes. If you know how the query optimizer works and the algorithms it uses, you know why (and when and where) it’s wrong. I think most people will find it easier to relate to a real life example, rather than discussing the internals of query optimizers, which is why I chose to link to this post.

And the problem is not isolated to MySQL, all the major database servers have these issues.

]]> By: jeremy http://labnotes.org/2008/08/23/dumb-and-faster-databases/comment-page-1/#comment-141219 jeremy Sun, 24 Aug 2008 21:29:38 +0000 http://blog.labnotes.org/?p=1175#comment-141219 Whenever I read something along the lines of "I tried to do xyz in SQL and I couldn't work it out/it wasn't fast enough, so instead I did it with zyx" my immediate thought is "I bet their schema, indexes or query is incorrect". I think this is definately the case here after reading the rest of the post and comments. In the worst case the product could be to blame (often MySQL), but another RDBMS will probably be able to handle what you're trying to do in a sensible fashion. Without looking at the alternatives you end up saying all RDBMS are bad for a particular task, whereas that isn't necessarily the case. Whenever I read something along the lines of “I tried to do xyz in SQL and I couldn’t work it out/it wasn’t fast enough, so instead I did it with zyx” my immediate thought is “I bet their schema, indexes or query is incorrect”. I think this is definately the case here after reading the rest of the post and comments.

In the worst case the product could be to blame (often MySQL), but another RDBMS will probably be able to handle what you’re trying to do in a sensible fashion. Without looking at the alternatives you end up saying all RDBMS are bad for a particular task, whereas that isn’t necessarily the case.

]]>