Question

So I’ve got a MySQL database that counts the number of servers with given usage levels on an hourly basis, as such:

Date                Status  Population
2012-01-13 15:33:40 UP      Standard
2012-01-13 15:33:40 UP      Light
2012-01-13 15:33:40 UP      Heavy
2012-01-13 15:33:40 UP      Light

2012-01-13 16:33:40 UP      Light
2012-01-13 16:33:40 UP      Very Heavy
2012-01-13 16:33:40 UP      Light

etc.

and a query that counts the number of servers with given usage levels grouped by minute, thusly

SELECT DATE, ROUND( UNIX_TIMESTAMP( DATE ) / ( 1*60 ) ) AS TIME, COUNT( IF( POPULATION =  "Light", 1, NULL ) ) AS LightCount, COUNT( IF( POPULATION =  "Standard", 1, NULL ) ) AS StandardCount, COUNT( IF( POPULATION =  "Heavy", 1, NULL ) ) AS HeavyCount, COUNT( IF( POPULATION =  "Very Heavy", 1, NULL ) ) AS VeryHeavyCount, COUNT( IF( POPULATION =  "Full", 1, NULL ) ) AS FullCount, COUNT( IF( POPULATION =  "Offline", 1, NULL ) ) AS OfflineCount
FROM  `Servers` 
GROUP BY TIME
ORDER BY DATE ASC

The output looks like:

DATE            TIME            LightCount  StandardCount   
2012-01-13 15:33:40 22108174    16          146     
2012-01-13 16:33:35 22108180    16          147 

and so on, with running counts for every hour.

I’m trying to find a way to determine the highest number of “LightCount” or “StandardCount” etc. within each 24 hour period. In other words, what was the highest hourly LightCount for any given day?

Is this at all possible? Would it require a nested query of some sort?

Thanks much for any help.

Answer 1

First, you can use GROUP BY DATE(DATE),HOUR(DATE) for hourly sums:

SELECT `DATE`, COUNT(...), ... -- HOUR(`DATE`) if you wanted to see it
FROM `Servers`
GROUP BY DATE(`DATE`),HOUR(`DATE`)
ORDER BY `DATE` ASC

Then to do maximum hourly counts per day you’d do

SELECT DATE(`DATE`) as Day, MAX(LightCount) as MaxHourlyLightCount, ...
FROM (
    SELECT `DATE`, ....
    FROM `Servers`
    GROUP BY DATE(`DATE`),HOUR(`DATE`)
    ORDER BY `DATE` ASC
     ) HourlyCounts
GROUP BY Day

Here’s a link to the MySQL Date/Time functions — you can do monthly stats with GROUP BY MONTH(``Date``), YEAR(``Date``), yearly stats with GROUP BY YEAR(``Date``), etc – very handy.

(Note: replace those double-backticks above with single-backticks — I don’t know how to escape them properly in stackoverflow).