CODEWITHSUNDEEP
phpmysql
zuk1
zuk1

Reputation: 18369

PHP MySql Storing Data for 24 hours

I'm building a toplist for my site and instead of recording in and out hits and emptying the database every 24 hours I would like to record in and out hits "within the last 24 hours", this way the toplist will always look busy no matter what time you look at it, while still acurately ranking sites.

So basically I'll have to add a record to the database logging an 'in' hit (for example) and the remove it around 24 hours later, or simply not consider it when ranking the sites.

I want to know the most efficient way of doing this, I have some ideas but they all seem really resource heavy. I hope you understand what I'm trying to achieve :) Any help would be appreciated.

Upvotes: 0

Views: 2226

Answers (5)

Argelbargel
Argelbargel

Reputation: 6210

If the time-constraint is not really hard (e.g. you'll loose money or are really annoying your users if the data is kept in the the db longer than 24 hours), I'd use use PHP's register_shutdown_function like this:


function cleanup() {
  // open db-connection etc.
  $query = 'DELETE FROM <yourtable> ' .
               'WHERE UNIX_TIMESTAMP(<timstampfield>) < ' . (time() - 86400);
  mysql_query($query);
  // close connection
}

register_shutdown_function('cleanup');

The above code assumes, <timestampfield> is of one of the the MYSQL-date-datatypes (TIMESTAMP, DATE, DATETIME).

Upvotes: 0

Vilx-
Vilx-

Reputation: 106920

I would delete the data that is older than 24 hours with a simple

DELETE...WHERE hit_time < now() - interval 24 hour

The other question is - when to call it? The tradeoff is between performance and stale data. The more often you call it, the less "stale" data there will be, but the server load will grow.

I see several approaches, pick one that suits your needs most:

  • Call it at the start of every script. This can be optmized by calling it only if the script will do something with the hit data. That way every script will always run with "correct" data. However this will have the maximum load.
  • Schedule a cron job and call it once every 1h/2h/24h/etc. This way there will be a little bit of "stale" data, but the overhead will be reduced to a minimum.
  • Do it like PHP does it with sessions - on every script startup give it a x% (x is configurable) chance of being run. That is, take a value from 0 to 100, and if it is smaller than x, execute the DELETE.

You can also invent some other schemes - for example, run it once per user session; or run it only if the execution time is evenly divisable by, say, 7; or something else. Either way you trade off performance for correctness.

Upvotes: 1

VolkerK
VolkerK

Reputation: 96159

You don't necessarily have to delete "old" data all the time. You can also limit the data set to records within the last 24 hours when querying data. 

WHERE
  site='xyz'
  AND hit_datetime > Now()-Interval 24 hour

see http://dev.mysql.com/doc/refman/5.1/en/date-and-time-functions.html

Upvotes: 3

Flo
Flo

Reputation: 1445

you could store the timestamp with each "hit" and then call a query like

 $time = time()-86400;
 mysql_query("DELETE FROM xxx WHERE timestamp < $time");

or you could same thing within the SELECT statement, depends on if you still need the hits afterwards, etc

Upvotes: 0

jao
jao

Reputation: 18610

Write a Stored Procedure that deletes records older than 24 hours. Then write a trigger that runs on every INSERT statement and calls the SP.

Upvotes: 0

Related Questions