Multi Relational Table Design Question

Hi! Im working on a webapplication and has serious thoughts about howto optimize my table structure. To explain:

My tablestructure today



(simplified):

tbl_customers
cust_id
name
.....

tbl_contacts
con_id
name
.....

tbl_groups
grp_id
name
.....


 
My subtables look like this(alternative 1):

tbl_sub_phone
phone_id
parent_type
parent_id
phone_area
phone_nr
.....

tbl_sub_email
mail_id
parent_type
parent_id
email
.....

 

As seen above every contact, group and customer can be assigned an unlimited amount of phonenumbers or emailadresses.
For example when entering a new email or a customer following will be inserted in tbl_sub_email: parent_type = 'cst', parent_id= '2' (the cust_id from tbl_customers), email = 'gwerg@fe.com'

The problem is i am uncertain if this is a very unefficient way of handling it? i see two alternatives:

Alternative 2:
i create x subtables for each table  for example tbl_customers will get its mailadresses and phonenumbers contained in tbl_customers_phone and tbl_customers_email
What i am uncertain of here is if this would make things alot more troublesome when searching på example after a specific phonenumber.

Alternative 3:




(simplified):

tbl_customers
cust_id
name
.....

tbl_contacts
con_id
name
.....

tbl_groups
grp_id
name
.....


 


tables connection objects to subobjects

tbl_customers_phone
id
cust_id
phone_id

tbl_contacts_phone
id
con_id
phone_id

tbl_customers_mail
id
cust_id
mail_id

 
subtables

tbl_sub_phone
phone_id
phone_area
phone_nr
.....

tbl_sub_email
mail_id
email
.....

 

Ranking these three models, wich would be the most efficient and most inefficient performanswise?
What i want to avoid is performanceproblems when listing the objects, my indexing skills are a bit limited although im doing alot of reading and testing regarding this.
So thats why im asking for advice so that i can minimize the need of rebuilding the table structure when the application already has been starting to get used.

I also have another general question.

I have alot of select querys when i need to fetch data from several different tables.
Most of them is that i for example get an application from tbl_applications table, and that tables contains the columns cat1, cat2 and cat3 (wich are categories and contain the primary key integer to the tbl_sub_categorys table)
With 3 joins i retrieve these 3 category names returning 1 result with all the info i need.

Since ive been getting som strange results from the query analyzer(i got results that using clustered indexing for the primary key resulted in a slower query (higher cost)) i actually have another question.

Can it generally be summed up that a single query(join or subquery) generaly ils faster than getting the data in separate selects?
In the example above this i have the options either of using joins = 1 query or doing 2 querys and sorting the categorys codewise in aspx pages or doing 4 querys, one for the app followed by 1 for every category.
Any input regarding this?

As i said earlier im looking for the most efficient way of doing the things abov, would greatly appriechiate any input!