Multi column primary key, which column to use as leading edge?

Given SQL SERVER 2000 and table:

  CityID int not null,
  PersonID bigint not null,
  SomeOtherAttributes null
WHERE CityID is a FK to Cities table,  and PersonID is only unique within a city (Off topic but FYI - the PersonId is imported from another application which does not use globally unique identifiers).

The primary key options are:

My question is which primary key to specify and why/how to figure out which primary key to specify?

Thank you!

Who is Participating?
lahousdenConnect With a Mentor Commented:
Both keys will serve well as PK.  However, there are some secondary considerations.  For example, if you see situations where selections are going to ask for all Person's in a particular City, or situations where reports on the Person table are to be ordered by CityID, then if you put CityId first in the Primary Key (like your first option) the query optimiser will be able to leverage the index that the Primary Key is based on, giving some performance benefits in these situations without requiring you to create an additional index on the table.
cityid,personid maybe of use...

but what is the cardinality of the two columns (how many values )

what do you normally search with city ? or person ?

to a cetain extent it may not matter unless you are intending to cluster in the key as well...

PersonID, CityId

As a general principle, you should have your most selective column first.
A proven path to a career in data science

At Springboard, we know how to get you a job in data science. With Springboard’s Data Science Career Track, you’ll master data science  with a curriculum built by industry experts. You’ll work on real projects, and get 1-on-1 mentorship from a data scientist.

mjmarlowAuthor Commented:
Some of my more complex queries  include a "CityId" qualifier in the join or where clause.   I noticed that  SQL Server index optimizer recommended then added an Index "CityId" on my Person table.  I wonder if I would spare the extra index overhead if i used CityId as the leading edge instead (of following the book of using most selective column).
That suggests that most of your queries are by CityId instead of PersonId. If that is the case, CityId would be the better option, as explained by lahousden
mjmarlowAuthor Commented:

Persons: 705151
Cities: 159

I often search for a subset  of (5-1000) persons who are isolated by City and presence in another table via a join on CityId and PersonId.

I do not intend on clustering the key, but from what i have read, clustering is done automatically if there are no other clustered indexes specified.
Question has a verified solution.

Are you are experiencing a similar issue? Get a personalized answer when you ask a related question.

Have a better answer? Share it in a comment.

All Courses

From novice to tech pro — start learning today.