Likes on posts on a heavy trafic site - php

Question

I have this website that lets people post stuff that other people can like, on all the main pages it shows a certain amounts of small tumbs with the amount of likes the post has. The website also has a 'hot' page wich shows the most liked post in the last 24 hours.
i curently was thinking about making the database like this:
CREATE TABLE likes (
`id` int(11) NOT NULL AUTO_INCREMENT,
`post` int(11) NOT NULL,
`user` int(11) NOT NULL,
`liked` TIMESTAMP DEFAULT CURRENT_TIMESTAMP,
PRIMARY KEY (`id`),
) ENGINE=MyISAM DEFAULT CHARSET=latin1 AUTO_INCREMENT=1 ;
but iam scared it will cause a lott of performace isues sinds the table will get really big really fast so to query it all it will slow my database down alott, would there be a good way to index this to help with performance? also would it be a good idea to make a second table just for the 'hot' page containing only the likes of the last 24hours wich i then take out the expired (older then 24hours) with a cron job every day?
im far from a expert on databases so some explination with the awnser would be much appreciated, thanks in advance

Answer 1

Your method would work if it is important to you to log who made which "likes". The table will grow linearly with use, which should not be a performance problem unless the site becomes very popular. Just be sure to use InnoDB tables as MyISAM will lock the entire table on writes and that is a write-heave database.
If simply knowing the amount of likes is sufficient, then do something like this:
CREATE TABLE IF NOT EXISTS posts (
`id` int(11) NOT NULL AUTO_INCREMENT,
`likes` int(11) NOT NULL,
PRIMARY KEY (`id`),
) ENGINE=InnoDB DEFAULT CHARSET utf8 AUTO_INCREMENT=1 ;
And on each "Like" run this query:
UPDATE posts SET `likes` = `likes`+1 WHERE id = {$id}";

Answer 2

Ok you can do one thing, First add the "Like" in your likes table and create another table naming "hotlikes" or whatever you want to call it....
Keep a field in hotlikes for DATE, and add date into it, automatically delete the dates which of previous date by creating a cron job
Show only 24hrs like from "hotlikes" table....

Answer 3

Lets say you have fairly large MySql(lets say aws ec2 large instance) which has been tuned properly, indexed properly, uses innodb as storage engine and uses caching you wouldn't have problem serving 100-200 simultaneous users. You will also have to consider what other tables or databases are on the same database machine is doing.
I suppose the post and user in the schema refers to the ids if yes then I would create the following indexes
Primary key (obviously)
post + timestamp
user + timestamp (I am guess you might want to query by users if not then this is not required)
Every 24 hours I will run script to build the top likes and store it in some kind of cache so that it can be retrieved with out hitting the db server wit optional write to db (fail safe just in case cache failed).
Another option is to use key-value stores(redis maybe) but them again it depends on whats yours use case and how much users you will have.

Answer 4

You can do what #Abhilash Shukla suggested you to, plus you can cache the number of likes on the post in the post table. You can update the cache once per X minutes, for example, like VBulletin and some other forums do for thread views. For this, you can create a table to store just id's of the posts, for which you need to recalculate the amount of likes. And do recalculation on cron.
You can even partition likes table by postid range, if you find that suitable for the task.