Re: Re-design of data storage table
Date: Mon, 2 Mar 2009 05:30:27 -0800 (PST)
On Mar 2, 4:39 am, __Paul <paulwragg2..._at_hotmail.com> wrote:
> Well I can see that has got quite interesting in the last few days.
> Erland - thank you for being sensible and realistic with your
> response, it is appreciated.
> I am not complaining that nobody else has come up with a better
> solution for me. Neither do I think that a few days of trawling the
> net without finding anything better means that there is not anything
> better. Believe me I have already put a lot of time into experimenting
> with different methods both on Oracle and SQL Sever. Also note that
> we're not all just bothered about improving our resume - some of us
> want to do a good job in the one that we are currently in, rather than
> already thinking about the next one.
> What I meant by trawling the net and finding no better solution is
> that all I have found is negative feedback, snide remarks and
> generally people looking down upon me because I have asked a question
> with regards to EAV. A few people have posted helpful suggestions in
> this thread, but others just want to have a laugh and a joke without
> actually suggesting a viable alternative. I don't know whether I
> phrase things incorrectly but I am not looking for anybody to do my
> job for me, just asking for a few simple pointers, advice etc.
> Up to now the advice has been run like hell, don't do it, ha ha you
> noob etc. Advice on an alternative approach would be more
> constructive. Or is the only alternative to have a table with as many
> columns as possible to support this?
Several times people have pointed out that scalable applications depend on a design that in some critical ways is based on the real world and entities/attributes relevant to the application. I don't think that this advice is off target based on what you have posted.
Perhaps some kind of hybrid approach where you do a fair amount of modeling then perhaps supplement the design with a small amount of on the fly attribute creation may scale ... perhaps. It depends a lot on how much load is introduced into the system and how much work it takes to build the SQL as well as to process the SQL generated.
In general I think many of the responders have posted relevant advice even if it may have a sour taste. I gotta admit words about trawling the net for a design did not sound good from my point of view.
What kind of testing and load validation approach are you planning exactly?
Trying to break your application and the database design is in many ways more important than you may realize. Received on Mon Mar 02 2009 - 07:30:27 CST