Please start any new threads on our new
site at https://forums.sqlteam.com. We've got lots of great SQL Server
experts to answer whatever question you can come up with.
| Author |
Topic |
|
exiledAussie
Starting Member
1 Post |
 Posted - 2011-11-16 : 14:31:39
|
| Hi,I am working on a c++ statistical analysis program that is to export data to relational databases using ODBC.We have a models which include domains and utilities, the latter specifying the value to the client of variables in the domain being in particular combinations.We would like to have a table that specifies these utilities, within a database that can include numerous different models, domains and utilities. But different utilities will be the function of different numbers of variables, so (it seems) we cannot utilise a generic table unless we include numerous, usually redundant, columns, which would be quite annoying to the end user. Alternatively, we can specify a new table whose name is based on the model - ie Model_X_Utility_1 - for each utility matrix. But this seems to be less that ideal.What would people suggest?EA |
|
|
visakh16
Very Important crosS Applying yaK Herder
52326 Posts |
Posted - 2011-11-17 : 00:40:22
|
| One option is to choose EAV model where you'll have generic table which stores utilities and variables along with their values as rows inside themhttp://en.wikipedia.org/wiki/Entity%E2%80%93attribute%E2%80%93value_modelAnother option if you're on SQL 2008 and above is Sparse columns which will still be different columns but will take very less storagehttp://blogs.msdn.com/b/sreekarm/archive/2009/01/08/sparse-columns-in-sql-server-2008.aspx------------------------------------------------------------------------------------------------------SQL Server MVPhttp://visakhm.blogspot.com/ |
 |
|
|
|
|
|
Variable Column Width Table or Alternative