Question

I have a table where we log users who have logged on to our system. It logs when they log on and when they log off. I need to produce some SQL which gives us a list of hours of the day and the number of people logged on in that time. I can do this for one time, but I do not know how to extend this to produce each hour of the day. I don’t really want to have to do 24 SQL statements if possible!?

SELECT     COUNT(userID) AS "count of users"
FROM         LogonTimes
WHERE     (LoginTime > '2011-09-12 09:00:00') AND (LogoffTime < '2011-09-12 10:00:00 ')

The above would produce say “3”. This shows 3 people were logged on between 9 and 10am.

Any ideas?!

[edit]It’s SQL Server as people have guessed below – sorry for not specifying! I’ll try out the suggestions and post back shortly! thanks 🙂 [/edit]

Answer 1

For SQL Server you could do…

declare @DayToCheck datetime 
set @DayToCheck = '2011-09-12'

;with C as
(
  select @DayToCheck as H
  union all
  select dateadd(hour, 1, H)
  from C
  where dateadd(hour, 1, H) < dateadd(day, 1, @DayToCheck)
)
select C.H as [hour],
       count(L.userID) as [count of users]
from C
  left outer join LogonTimes as L
    on L.LogoffTime > C.H and 
       L.LoginTime < dateadd(hour, 1, C.H)
group by C.H       

Try here: https://data.stackexchange.com/stackoverflow/q/112462/

A version for SQL Server 2000 that uses a number table instead if a CTE. Here I use master..spt_values.

declare @DayToCheck datetime
set @DayToCheck = '2011-09-12'

select dateadd(hour, N.number, @DayToCheck) as [hour],
       count(L.userID) as [count of users]
from master..spt_values as N
  left outer join LogonTimes as L
    on L.LogoffTime > dateadd(hour, N.number, @DayToCheck) and
       L.LoginTime < dateadd(hour, N.number + 1, @DayToCheck)
where N.Type = 'P' and
      N.Number between 0 and 23       
group by dateadd(hour, N.number, @DayToCheck)