MySQL - Large database, improve search time - mysql

Question

I have two rather big tables (threads and posts) that include a ton of forum posts. I really have to improve my search time. Even doing a normal search where COLUMN = VALUE will take 15 seconds. Doing a LIKE often crash the entire website (timeout).
Here's a picture of my site and two tables:
The threads table contains about 430,000 rows.
The posts table contains about 2,700,000 rows.
And I need to combine these in a query to get the results I want.
Don't bother about the search boxes on the website for now. Let's just start off with this query right here and start improving this one first.
SELECT p.id, t.id, t.title, t.threadstarter, t.replies, t.views, t.board, p.dateposted FROM threads t
JOIN posts p
ON t.id = p.threadid
WHERE t.title = 'sell'
GROUP BY t.id
This query will take about 15 seconds to get all threads and posts where the thread title is "sell". How would I improve this, making it just a second or two? Is this even possible with MySQL in two tables with these sizes?
And from there on, I would have to make a LIKE (unless there is another method). Because the users on the website will most likely not search for an exact match. And I'd want to include any title that includes the world "sell". So that would be like this:
SELECT p.id, t.id, t.title, t.threadstarter, t.replies, t.views, t.board, p.dateposted FROM threads t
JOIN posts p
ON t.id = p.threadid
WHERE t.title LIKE '%sell%'
GROUP BY t.id
Which I am not even going to bother measuring. It's crashing the website (too long time to execute). So this one really(!) needs improvement.
How should I even approach this? Should I even use MySQL? What options do I have? I do not want a user to sit and wait 30-300 seconds for a query to finish. At most 5 seconds.
Is this possible, with such large tables?
I've heard using "MATCH" and "AGAINST" could be better than a "COLUMN" LIKE "VALUE". But then I need to make all the columns freetext. Any downsides of doing that?
If there's anyone out there that's worked with a ~3 million row MySQL database, then please let me know how you handled it (if you did).

Answer 1

Make use of INDEX. Just try to create an index on one of the table which has more records or the master though its an inner join still itll make it easier to inner join the two.
Plus, I simply din understand usage of group by without any aggregation as its select *.. in the query.
CREATE INDEX Index_NAME ON
threads(title);

Answer 2

The correct way to express your first query is:
SELECT p.id, t.id, t.title, t.threadstarter, t.replies, t.views, t.board, p.dateposted
FROM threads t JOIN
posts p
ON t.id = p.threadid
WHERE t.title = 'sell' AND
p.dateposted = (SELECT MIN(p2.dateposted) FROM posts p2 WHERE p2.threadid = p.threadid);
This gets rid of the GROUP BY, so it might improve performance. In particular, you want indexes on:
threads(title, id)
posts(threadid, dateposted)

Answer 3

give these two articles a read.
how to optimize mysql queries for speed and performance
MySQL Optimization

Answer 4

LIKE with a leading wild card must scan all 430,000 rows:
WHERE t.title LIKE '%sell%'
Change to this:
WHERE MATCH(t.title) AGAINST('+sell' IN BOOLEAN MODE)
and have
FULLTEXT(title)
With that setup, the query can go directly to the few rows that have the 'word' sell in it.
Caveat: There are restrictions on what FULLTEXT can search for -- only "words', not "stop words", only words of a certain minimum length, etc.