AutoNumber Question

[accessquestions]

Hello,
As far as a Primary Key is concerned (for an Employee Table)
I have read numerous times on the boards NOT to use the Auto Number feature,
I'm not sure exactly why, although I could see where the random number can
confuse some.
Would anyone recommend using a SSN instead? It will be unique, and it will
serve as a good reference to the employee.
Any thoughts of why or why not?

Thanks Again !

VL

 

[tina]

well, personally, i would not use the SSN as a primary key, for 2 reasons.
first, SSNs are unfortunately not unique; there are many, many duplicates
floating around out there. of course, only one person is using a specific
SSN *legally*, but unless you want to get into tracking down fraudulent
use... second, if you use the SSN as primary key, you'll have it populating
multiple tables, potentially scattered all over your database. again,
personally, i would safeguard an SSN the same way i would a credit card
number, in my database; put it in one place, and one place only, so it's
easier to protect by whatever means you use to secure proprietary data.

having said all that, there's no reason you can't use an Autonumber as the
primary key field for your employee table *as long as it is a value the has
no meaning to anyone, only to Access*. your user should never see an
Autonumber primary key, or use it directly to search for a record in the
table. don't confuse how your *user* finds a certain record in a table with
how *Access* identifies each record in a table. you can easily use an
Autonumber primary key and still allow your user to seach for an employee
record using the person's SSN.

hth

 

[accessquestions]

good point.
the Auto Number is sort like a VIN on a car, you rarely have to see it, but
its there and its serving a purpose,, and nobody committs it to memory, if
they would it would just confuse them. you convinced me, im an auto number
guy from now on.
thanks

 

[tina]

well, glad i could help.

but i have to point out that the VIN analogy is not apt. a VIN is not just a
collections of numbers/letters, it is a very specific code that identifies
manufacturer, make and model of vehicle, model year of vehicle, sometimes
where the vehicle was built, sometimes engine size, sometimes maximum
payload for trucks, and other data - as well as the last 4 to 6 digits that
are unique to a specific vehicle that is otherwise identical to many other
vehicles in all those other respects. so a VIN has meaning separate from any
tracking database that stores it, and it must often be seen and used
directly by people - such as manufacturers, owners, insurers, DMVs.

an Autonumber primary key, on the other hand, has no meaning outside of the
database where the data is stored, and no meaning within the database,
except to the software - and it absolutely should *not* be used with the
idea of attaching a meaning to it. it also should not be seen by anyone who
uses the database, at any time - excepting only that the developer may look
at Autonumber values from time to time while building and/or troubleshooting
the database.

hth

 

[accessquestions]

understood. I actually thought of that after I posted.
Just goes to show that it really takes an analytical mind to perfect the
concepts of relational databases.
A VIN, like you state, serves a purpose, and is an attribute (an important
one) of a vehicle. Not just a random number to identify it's uniqueness in
a database. If it were a Database containing vehicle information, the VIN
would not be a proper number to use a primary key identifier either (just
like a SSN), but instead a value relating to another value. And even then,
a random number, such as Aceess' AutoNumber should be used for the system to
identify it and its uniqueness.
This dialogue was very educational for me as it relates to what I have read
many times on the boards ut it all on paper first", and, "determine all
information to be stored in the database and its relations to other tables"
before creating them. If this was done, I would know that the SSN and VIN
Numbers were not suitable candidates for a Unique Identifier.
Point Taken !
Thanks for all your insight ,as usual

VL

 

[tina]

Just goes to show that it really takes an analytical mind to perfect the

concepts of relational databases.

This dialogue was very educational for me as it relates to what I have read
many times on the boards ut it all on paper first", and, "determine all
information to be stored in the database and its relations to other tables"
before creating them. If this was done, I would know that the SSN and VIN
Numbers were not suitable candidates for a Unique Identifier.

then it was time well spent, for both of us. and hopefully the dialogue be
helpful to other folks who may read this thread.

Thanks for all your insight ,as usual

and you're very welcome, as always.

 

[BruceM]

I would like to point out that a primary key does not need to be meaningless
to the user. A sequential number such as 000001, 000002, etc. may be used
for invoice numbers or the like. I don't see any reason why that number
can't be the PK.
There are legal reasons why SSN should not be widely available in a
database, as it would be if it was part of relationships. However, I would
think an employer would want to know if the number is not unique. If John
Smith leaves the company, and five years later Jill Jones shows up with John
Smith's SSN, I expect it would be a real mess for a company that simply
ignores the duplication.
A VIN as a PK could get a bit murky for an insurance company or the DMV if a
vehicle changes hands, but I expect either entity would want to know if a
VIN is being reused. A VIN could make sense as a PK in some situations.
Uniqueness is one thing; suitablility for use as a PK is another. There are
a lot of possible circumstances, each of which needs to be considered
separately.

 

[tina]

you're quite right, Bruce - a primary key does not need to meaningless to
the user. in this thread we were discussing specifically the *Autonumber* as
primary key, so the posted comments were applicable. i agree with you also
that a unique value is not necessarily a good candidate for a natural key in
a table; that decision should be made on a table by table basis in every
relational data model, based on a number of factors that may or may not be
unique to that business process.

 

[BruceM]

Hi Tina,

I could have made it clearer that I was responding specifically to the
following:

"If it were a Database containing vehicle information, the VIN would not be
a proper number to use a primary key identifier either (just like a SSN),
but instead a value relating to another value. And even then, a random
number, such as Access' AutoNumber should be used for the system to identify
it and its uniqueness."

I just wanted to point out that there is no need to get stuck on using such
a number. To say that something such as as autonumber *should* be used as
the PK is to create a rule where one is not needed.

 

[tina]

again, i quite agree. even in cases where no natural key is available in a
table, Autonumber is not the only alternative, and certainly not the best
alternative in *every* situation.

 

[accessquestions]

Good point. But does this mean that setting the No Duplicates option on a
particular field only applies if that field is a Primary Key? I was under
the impression that it you can select No Dups for any Field. So, with the
SSN or VIN anaolgies, couldnt one still ensure the uniqueness of them, even
if they werent the PriKey? And still use the AutoNumber option for the
Primary?

 

[tina]

comments inline.

Good point. But does this mean that setting the No Duplicates option on a
particular field only applies if that field is a Primary Key?
no.

I was under
the impression that it you can select No Dups for any Field.

that is correct.

So, with the
SSN or VIN anaolgies, couldnt one still ensure the uniqueness of them, even
if they werent the PriKey?

well, to be more precise, you can set a unique index on any field to ensure
that no duplicate values can be entered in the field. you may have duplicate
*data*, but you can't enter that duplicate data in a field with a Unique
index setting.

 

[accessquestions]

alrighty then,
thanks Tina!

comments inline.



that is correct.


well, to be more precise, you can set a unique index on any field to
ensure
that no duplicate values can be entered in the field. you may have
duplicate
*data*, but you can't enter that duplicate data in a field with a Unique
index setting.

 

[tina]

you're welcome

alrighty then,
thanks Tina!