Question

Lets say I have 2 tables. The first table, table_1, contains each posted content including

table_1
   title, 
   author name, 
   email, 
   city_name, etc. 

The second table provides a lookup for table_1. It has 2 columns,

table_2
   city_id and 
   city_name. 

For instance, city_id =1 corresponds to New York, city_id =2 corresponds to Chicago… and so on. Under the ‘city’ column in table1, the city_id is listed which can easily be joined with table 2, producing a readable city name.

Would the following statement be as efficient as using a WHERE with city_id? Reason being is that I would be filtering results based on a city which is a string and I don’t want (or need?) to correlate each input to its matching ID number in table2.

SELECT table1.city, table2.city_name 
FROM table1 
WHERE table2.city_name=(input city name)
JOIN table2.city_name ON table2.city_id = table1.city

Answer 1

Because the join is an inner join the following should lead to equivalent execution plans. (That is, they should exhibit the same performance characteristics — write the SQL clearly and let the SQL engine do the dirty optimization work.)

As presented in the other answers:

SELECT table1.*, table2.city_name 
FROM table1 
JOIN table2 ON table1.city_id = table2.city_id
WHERE table2.city_name = (city_input);

And, as what I believe is the “optimized form” presented in the question:

SELECT table1.*, t2.city_name 
FROM table1 
JOIN (SELECT * FROM table2
      WHERE table2.city_name = (city_input)) AS t2
ON table1.city_id = t2.city_id

This is because of the Relation Algebra Model that SQL follows; at least under RA the equality (“selection”) here can be moved across the join (a “natural join” in RA) while keeping the same semantics. Of course, “to make for certain”, run a query analysis. The most basic form is using EXPLAIN.

Happy coding.