低效的MySQL数据库/语句降低了PHP / JS系统的运行速度

Question

I have developed a fairly ramshackle PHP/JavaScript Reward system for my school's VLE.

The main bulk of the work is done on a transactions table which has the following fields:

Transaction_ID , Datetime , Giver_ID , Recipient_ID , Points , Category_ID , Reason

The idea is that if I, as a member of staff, give a student some Reward Points, an entry such as this is inserted into the database:

INSERT INTO `transactions` (`Transaction_ID`, `Datetime`, `Giver_ID`, `Recipient_ID`, `Points`, `Category_ID`, `Reason`) VALUES
(60332, '2012-02-22', 34985, 137426, 5, 5, 'Excellent volcano homework.');

This worked fine - but I didn't really consider just how much the system would be used. I now have over 72,000 transactions in this table.

As such, a few of my pages are starting to slow down. For instance, when staff try to allocate points, my system runs a command to get the member of staff's total point allocation and other snippets of information. This appears to be displaying rather slowly, and looking at the MySQL statement/accompanying PHP code, I think this could be much more efficient.

function getLimitsAndTotals($User_ID) {

    $return["SpentTotal"] = 0;
    $return["SpentWeekly"] = 0;

    $sql = "SELECT * 
        FROM `transactions` 
        WHERE `Giver_ID` =$User_ID";
    $res = mysql_query($sql);

    if (mysql_num_rows($res) > 0) {
        while ($row = mysql_fetch_assoc($res)) {
            $return["SpentTotal"] += $row["Points"];

            $transaction_date = strtotime ($row["Datetime"]);

            if ($transaction_date > strtotime( "last sunday" )) {
                $return["SpentWeekly"] += $row["Points"];
            }
        }
    }

    return $return;
}

As such, my question is twofold.

1. Can this specific code be optimised?
2. Can I employ any database techniques - full text indexing or the like - to further optimise my system?

---

EDIT : RE Indexing

I don't know anything about indexing, but it looks like my transactions table does actually have an index in place?

Is this the correct type of index?

Here is the code for table-creation:

CREATE TABLE IF NOT EXISTS `transactions` (
  `Transaction_ID` int(9) NOT NULL auto_increment,
  `Datetime` date NOT NULL,
  `Giver_ID` int(9) NOT NULL,
  `Recipient_ID` int(9) NOT NULL,
  `Points` int(4) NOT NULL,
  `Category_ID` int(3) NOT NULL,
  `Reason` text NOT NULL,
  PRIMARY KEY  (`Transaction_ID`)
) ENGINE=InnoDB  DEFAULT CHARSET=latin1 AUTO_INCREMENT=74927 ;

---

Thanks in advance,

Answer 1

Make sure Giver_ID is indexed . Try also running the strtotime outside of your while loop as I imagine its an expensive operation to be running 72,000 times.

if (mysql_num_rows($res) > 0) {

    // assign the unix timestamp of last sunday here.
    $last_sunday = strtotime('last sunday');

    while ($row = mysql_fetch_assoc($res)) {
        $return["SpentTotal"] += $row["Points"];

        $transaction_date = strtotime ($row["Datetime"]);

        if ($transaction_date > $last_sunday) {
            $return["SpentWeekly"] += $row["Points"];
        }
    }
}

Also consider running UNIX_TIMESTAMP(Datetime) AS datetime_timestamp in your SQL instead of getting it out as a string and running another expensive strtotime operation. You can then simply run:

if (mysql_num_rows($res) > 0) {

    // assign the unix timestamp of last sunday here.
    $last_sunday = strtotime('last sunday');

    while ($row = mysql_fetch_assoc($res)) {
        $return["SpentTotal"] += $row["Points"];

        if ($row['datetime_timestamp'] > $last_sunday) {
            $return["SpentWeekly"] += $row["Points"];
        }
    }
}

(if your column is of type DATE, of course!)

Answer 2

72k is nothing... However there could be other things here that are causing slowdowns, like your mysql configuration (how much memory you allocated to the service), and a few others (do a search for optimizing mysql).

Also look at how many INSERTS you have in a given page, that will typically slow down your page load, if you are doing an insert every action for a user, you are making a costly transaction.

Typically a SELECT is much less expensive than an INSERT, etc;

From your SQL code I can also assume you didn't optimize your queries, something like:

$sql = "SELECT * 
    FROM `transactions` 
    WHERE `Giver_ID` =$User_ID";

Should be:

$sql = "SELECT Points, DateTime FROM ..."

(not a big issue from your data, but it only requests WHAT YOU NEED, not everything which would require more memory while going through the iteration).

Also not sure how your schema is designed, but are you using indexing on Transaction_ID ?

Answer 3

Can this specific code be optimized?

Yes.
But definitively not by "looking at the code".

Can I employ any database techniques - full text indexing or the like - to further optimise my system?

Nope, you can't.
Simply because you have no idea, what certainly is slow in your system.

This is merely a common sense, widely used in the real life by anyone, but for some reason it become completely forgotten when someone start to program a web-site.

Your question is like "my car is getting slow. How to speed it up?". WHO ON THE EARTH CAN ANSWER IT, knowing no reason of the slowness? Is it tyres? Or gasoline? Or there is no road around but an open field? Or just someone forgot to release a hand brake?

You have to determine the slowness reason first.

Before asking a question you have to measure your query runtime.
And if it's still fast, you should not blame it for the slowness of your site.
And start profiling it.

First of all you have to determine what part of the whole page makes it load slow. It could be just some Javascript loading from the third-party server.
So, start with "Net" tab in the Firebug and see the most slow part.

then proceed to it.
If it's your PHP code - use microtime(1); to measure different parts to spot the problem one.

Then come with the question.

However, if take your question not as a request for the speed optimization, but for the more sanity in SQL, there are some improvements that can be made.

Both numbers can be retrieved as a single values.

select sum(Points) as Total from FROM `transactions` WHERE Giver_ID=$User_ID

will give you tital points
as well as

$sunday = date("Y-m-d",strtotime('last sunday'));

select sum(Points) as Spent from FROM `transactions` 
WHERE Giver_ID=$User_ID AND Datetime > '$sunday'

will give you weekly points.