how to optimize my query? - php

Question

I have 3 tables country_data, user_data and topic_data with table structures as given.
country_data:
name | code
---------------|---------------
India | IN
United States | US
Australia | AU
user_data:
user_ip | topic_code | country
---------------|---------------|---------------
192.168.1.1 | topic_code_1 | India
192.168.1.2 | topic_code_2 | United States
192.168.1.3 | topic_code_3 | Australia
topic_data:
name | code
---------------|---------------
topic_1 | topic_code_1
topic_2 | topic_code_2
topic_3 | topic_code_3
I have about one hundred thousand(100,000) rows in user_data table.
What I want is, I need to filter the count of users from each country with its corresponding country code for a given topic. For example, I need the count of users who viewed topic_2 in each country. The requered output format is
country_code | count
---------------|---------------
IN | 150
US | 120
AU | 100
Now please check my query:
SELECT cd.code, COUNT(ud.country) as count
FROM topic_data as td, user_data as ud, country_data as cd
WHERE td.name = 'topic_1' AND td.code = ud.topic_code AND ud.country = cd.name
GROUP BY ud.country
This one takes about 2 seconds to complete the execution in phpmyadmin. In the php webpage, it takes 15 seconds to load the page even in the server. by removing the group by in the query, ie GROUP BY ud.country, it takes more than 30 seconds to execute and the output is with the last country code and total of all countries visits. what am I doing wrong? please help.
----UPDATE----
Altered the tables using foreign keys and so as my queries too. now it works with lightning speed. thanks for those who helped.

Answer 1

The query doesn't look too bad IMO. However the normalization of the data looks a bit strange, e.g. why would you have a country (name) field on user_data table, just to join into country on name to look up the code? Instead, the more logical thing to me would be to reference country by country code (or other indexed key constraint). This would also save a join to country, if you just need the code as per your example query. If user_data is a high volume table, you will want to keep the data in it to a minimum to reduce IO when reading (density).
Also, as an aside, joining using JOIN instead of in the WHERE clause will improve the readability of your code, IMO:
SELECT cd.code, COUNT(ud.country) as count
FROM topic_data as td
INNER JOIN user_data as ud
ON td.code = ud.topic_code
INNER JOIN country_data as cd
ON ud.country = cd.name
WHERE td.name = 'topic_1'
GROUP BY ud.country;
To address the performance issues, check that the following indexes are in place:
Index on topic_data.name
Index on the foreign keys user_data.topic_code and user_data.country (or user_data.country_code if you do change the foreign key to user_data.country_code)

Answer 2

try this instead:
use below database structure for using numerical matching in INNER JOIN statement may decrease search time,
so index your id column of tables (e.g. primary key):
**country_data**
id|name | code
--|---------------|---------------
1 |India | IN
2 |United States | US
3 |Australia | AU
**user_data**
user_ip | topic_id | county_id
---------------|-----------|---------------
192.168.1.1 | 1 | 1
192.168.1.2 | 2 | 2
192.168.1.3 | 3 | 3
**topic_data**
id|name
--|------------
1 |topic_1
2 |topic_2
3 |topic_3
and run multiple INNER JOIN statment like:
SELECT cd.code, count(ud.topic_code) as count
FROM ud
INNER JOIN cd ON cd.id = ud.country
INNER JOIN td ON td.id = ud.topic_code
WHERE td.code='topic_1'
GROUP BY ud.country;