Design Consideration for Very Wide and Very Deep Table

pask_HN

We are a eCommerce retail business and are debating how to handle the addition of 25 new columns to an already wide (125 columns) and deep (21 million) table (email/customer).  The new fields won't be used by as many people as some of the existing.  We are debating between adding the 25 to the email table, or creating a new table and creating a snowflake in the RPD, joining on email ID.  This "snowflake" would be atypical in that it is a 0,1:1 relationship.

I'd appreciate any insights given!

Thank you.

[Deleted User]

Not necessarily an untypical situation.

It's hard to tell you "Do A" or "Do B" since we know nothing about your database or the physical modelization. There are 50'001 possibilities in the DB to make this perform well....or not. Have you run any tests so far? Do you have any DBAs who know something about performance tuning with regards to storage and physical modeling?

There are so many options and moving things that anybody giving an absolute answer - is simply loudly voicing the option he thinks is correct.