I have a query that will show the past 3 salaries within a 2 year period (if there are three) I have the query up and running the problem is, is its extremely slow... I'm wondering if there was a better way to write this query. I'm some-what new to oracle.
Here are my Tables
TABLE 1: Salary
ASSIGN_ID | start_date | end_date | salary |
1 | 11/27/2017 | 1/05/2018 | 50000.0 |
2 | 1/06/2018 | 6/08/2018 | 76000.0 |
3 | 6/09/2018 | 12/31/4712 | 80500.0 |
TABLE 2: Assignments
ASSIGN_ID | per_ID | start_date | end_date |
1 | 1 | 11/2/2017 | 1/05/2018 |
2 | 1 | 1/06/2018 | 6/08/2018 |
3 | 1 | 6/09/2018 | 12/31/4712 |
4 | 2 | 5/12/2016 | 7/18/2017 |
5 | 2 | 7/19/2017 | 12/31/4712 |
Table 3: Person
per_id | first_name | last_name |
1 | John | Smith |
2 | Jane | Doe |
Our end dates default to 12/31/4712 if they're are currently active in the assignment
My Query looks like this:
SELECT
per.first_name,
per.last_name,
(CASE WHEN sal1.start_date >= add_months(CURRENT_DATE, -24)
THEN sal1.salary
ELSE NULL END) oldest_salary,
(CASE WHEN sal2.start_date >= add_months(CURRENT_DATE, -24)
THEN sal2.salary
ELSE NULL END) prior_salary,
sal3.salary current_salary,
FROM
person per
INNER JOIN assignments asg1 ON asg1.per_id = per.per_id
INNER JOIN assignments asg2 ON asg2.per_id = asg1.per_id
INNER JOIN assignments asg3 ON asg3.per_id = asg2.per_id
INNER JOIN salary sal1 ON sal1.assign_id = asg1.assign_id
INNER JOIN salary sal2 ON sal2.assign_id = asg2.assign_id
INNER JOIN salary sal3 ON sal3.assign_id = asg3.assign_id
WHERE asg3.start_date =
(SELECT MAX(asg.start_date
FROM assignments asg
WHERE asg.assign_id = asg3.assign_id)
AND (asg3.start_date - 1) BETWEEN asg2.start_date and asg2.end_date
AND (asg2.start_date - 1) BETWEEN asg1.start_date and asg1.end_date
AND sal1.salary != sal2.salary
AND sal2.salary != sal3.salary
ORDER BY 2,1
Is there a simpler way to do this? because when I run my script it processes forever. I think I might need better joins. like I said i'm new and my understanding of joins is weak.
A simpler form:
SELECT
z.first_name,
z.last_name,
--typical cross-db compatible pivot method
MAX(CASE WHEN z.rown = 1 THEN z.salary END) as recentsalary,
MAX(CASE WHEN z.rown = 2 THEN z.salary END) as oldersalary,
MAX(CASE WHEN z.rown = 3 THEN z.salary END) as oldestsalary
FROM
(
SELECT
per.first_name,
per.last_name,
--number assignments from 1=recent to N older
row_number() over(partition by a.per_id order by a.start_date desc) rown
s.salary
FROM --join up all
person p
INNER JOIN assignments a ON a.per_id = p.per_id
INNER JOIN salary s ON s.assign_id = s.assign_id
WHERE a.end_date > ADD_MONTHS(SYSDATE, -36) --only recent 3 years
) z
WHERE z.rown <= 3 --only the most recent 3 assignments
GROUP BY first_name, last_name --achieve pivot
It works by:
Join up all data so people, assignments and salaries are known
Only consider assignments ended since 3 years ago
Number the assignments in youngest to oldest order (1=youngest)
Pivot the top 3 numberings into 3 columns for recent, older and oldest salary, per person