Project takes long time to load - slow loading

Sometimes when you have installed a new project or you are opening the project from a different computer you may experience that the loading of the project is notably slow.

There are several possible causes for this problem.

Possible causes

Old version of Acinta Intelligence Suite

Check if the version of Intelligence Suite you are trying to open is up to date. If metadata was edited using a newer version then it may have stored properties that are not understood by older versions. When this happens, an error trapping mechanism is invoked that writes all the unrecognized properties to the log file. You can verify the problem by investigating the log file.

SOLUTION  Update the .exe files to a new version.

Upgrade to Acinta Intelligence Suite 2014

If you recently upgraded to Acinta Intelligence Suite 2014 from an older version then the problem may be caused by obsolete properties in metadata. I.e. properties that are no longer supported by the new version. Such properties cause internal validation errors and a lot of writing to the log-files.

SOLUTION  You must "clean" the metadata by loading the project in Modeller and run the "Repair" utility and then save the project. It is best to use the option "Save all objects" but it may take somewhat longer to process.

Slow network

A lot of metadata must be transferred over the network when Datashop, Modeller etc. opens a project. A slow network may cause this process to take a very long time. Note: Look out for wireless networks since these are notably slower than wired networks.

SOLUTION  Wire the PC to a cable or upgrade/fix problems in the (wireless) network.

Too little RAM

If the server has too little RAM to fit all the running processes inside then some memory will be mapped to disk. Since disks are multitudes slower than RAM this will inevitably lead to prolonged load times during start-up because thousands of dynamic objects are created representing metrics, dimensions, dashboards etc. This problem is typical of virtual machines where RAM is often a scarce resource to be shared across many virtual machines.

SOLUTION  Enhance the server with more RAM. If the server is a virtual machine it should be considered whether the RAM is dedicated to the server or dynamically shared. Dedicated RAM is better.

Antivirus system interference

Antivirus software such as Trend Micro may cause the project to load extremely slow due to dynamic scanning of running processes.

SOLUTION  Turn off the antivirus system or the parts hereof that cause the problem, e.g. "process scanning". Or perhaps better: Add the Acinta applications (Datashop, Modeller etc.) to the antivirus' so-called White-list to prevent the AV software from scanning these applications at run-time.

Fiscal/financial year date calculations

As a followup to my previous article on calculation of dynamic dates I will demonstrate how we can make similar calculations that take the company's financial year into consideration.

This is important for companies that have a financial year that is not coinciding with the calendar year. In Denmark the financial year typically starts at the beginning of a quarter, i.e.:

• 1st of January
• 1st of April
• 1st of July, or
• 1st of October

Thus, if in our dashboard we want to use a "year-to-date" filter then we actually want the filter to span from the beginning of the financial year till today's date.

Such a filter condition can be implemented using this SQL:

DATEADD(dd, 0, DATEADD(mm, 6, DATEADD(yy, CASE WHEN MONTH(getdate()) >= 7 THEN YEAR(getdate()) ELSE YEAR(getdate()) - 1 END - 1900, 0)))

The result of this expression (assuming today's date is January 13th 2012) is:

'2011-07-01'

i.e. an ISO formatted date, which should be valid in any SQL Server.

The above example assumes that the financial year starts on July 1st, but obviously you can use the approach for any start day of the financial year.

To change the starting month of the financial year you simple have to modify the script in two places:

• 6: Replace this value with the last month in the financial year
• 7: Replace this value with the first month in the financial year

Filter modifiers
As usual, date functions are very useful with filter modifiers. Using a filter modifier you can implement a report where the user selects a date and then your filter modifier transforms the selection into a "year-to-date" date range using the user's selection.

To use the above script in a filter modifier you simply have to replace the occurences of "getdate()" with the desired filter modifier placeholder, {Value1} or {Value2}.

I.e. you need to use this variation of the SQL expression:

DATEADD(dd, 0, DATEADD(mm, 6, DATEADD(yy, CASE WHEN MONTH({Value1}) >= 7 THEN YEAR({Value1}) ELSE YEAR({Value1}) - 1 END - 1900, 0)))

How to calculate dynamic dates in SQL Server

Quite often we need to calculate some kind of dynamic date based on the current date or the user's selection in a date filter. Examples include "First day of the year", "First day of the month" and so on.

To help you build such date formulas I have assembled a set of useful calculations for you to use.

Note: To use the calculations in Filter Modifiers you simply replace "getdate()" with the desired filter placeholder, e.g. "{Value1}".

First Day of Month

DATEADD(mm, DATEDIFF(mm,0,getdate()), 0)

Monday of the Current Week

DATEADD(wk, DATEDIFF(wk,0,getdate()), 0)

First Day of the Year

DATEADD(yy, DATEDIFF(yy,0,getdate()), 0)

First Day of the Quarter

DATEADD(qq, DATEDIFF(qq,0,getdate()), 0)

Midnight for the Current Day

DATEADD(dd, DATEDIFF(dd,0,getdate()), 0)

Last Day of Prior Month

DATEADD(dd,-1,DATEADD(mm, DATEDIFF(mm,0,getdate() ), 0))

Last Day of Prior Year

DATEADD(dd,-1,DATEADD(yy, DATEDIFF(yy,0,getdate() ), 0))

Last Day of Current Month

DATEADD(dd,-1,DATEADD(mm, DATEDIFF(m,0,getdate() )+1, 0))

Last Day of Current Year

DATEADD(dd,-1,DATEADD(yy, DATEDIFF(yy,0,getdate() )+1, 0))

First Monday of the Month

DATEADD(wk, DATEDIFF(wk,0, dateadd(dd,6-datepart(day,getdate()),getdate())), 0)

First Day of the Month 12 Months Ago
DATEADD(mm, DATEDIFF(mm,0,getdate()) - 12, 0)
Note: You can change "12" to any number of months ago you wish.

Fejl: "Repository databases are not correctly configured" - problem med danske tegn

Hvis du får beskeden:

---------------------------
Warning
---------------------------
Repository databases are not correctly configured.
You may re-configure repository from menu: File|Configure repository.

når du åbner Datashop/Modeller, så skyldes det muligvis problemer med tegnsæt og ukendte tegn i Projects.dspx filen.


Åbn din Projects.dspx fil og kontrollér den for danske bogstaver og andre specialtegn og fjern dem.


En anden mulighed - men dette har jeg dog ikke testet - er at gemme filen i utf8 Unicode format.

ISO ugenumre i AcTimeDim - rettelse til fejl i ugenumre

Nogle ældre versioner af metadata havde desværre fejl i ugenumrene.
Det er let at rette ugenumrene til korrekte ISO ugenumre, dvs. de ugenumre, der anvendes bredt i Europa med nedenstående scripts:


For at rette ugenumrene skal du gøre flg.:

1.       Åbn SQL Server Management Studio
2.       Åbn en query
3.       Vælg databasen ”AcintaXXMeta” (NN er systemnavnet)
4.       Indsæt nedenstående SQL og eksekvér det

create function dbo.F_ISO_WEEK_OF_YEAR
  (
  @Date      datetime
  )
returns int
as
/*
Function F_ISO_WEEK_OF_YEAR returns the
ISO 8601 week of the year for the date passed.
*/
begin
declare @WeekOfYear int
select
  -- Compute week of year as (days since start of year/7)+1
  -- Division by 7 gives whole weeks since start of year.
  -- Adding 1 starts week number at 1, instead of zero.
  @WeekOfYear =
  (datediff(dd,  case  -- Case finds start of year
    when NextYrStart <= @date then NextYrStart
    when CurrYrStart <= @date then CurrYrStart
    else PriorYrStart
  end,@date)/7)+1
from
  (
  select
  -- First day of first week of prior year
    PriorYrStart =
      dateadd(dd, (datediff(dd,-53690,dateadd(yy,-1,aa.Jan4))/7)*7,-53690),
  -- First day of first week of current year
    CurrYrStart =
      dateadd(dd,(datediff(dd,-53690,aa.Jan4)/7)*7,-53690),
  -- First day of first week of next year
    NextYrStart =
      dateadd(dd,(datediff(dd,-53690,dateadd(yy,1,aa.Jan4))/7)*7,-53690)
    from
    (
       select
       --Find Jan 4 for the year of the input date
         Jan4 =
dateadd(dd,3,dateadd(yy,datediff(yy,0,@date),0))
    ) aa
  ) a
return @WeekOfYear
end

go

UPDATE AcTimeDim
SET CalendarWeek = CAST(CASE WHEN (dbo.F_ISO_WEEK_OF_YEAR(AcDate) > 50) AND(Month(AcDate) = 1) THEN YEAR(AcDate) - 1 ELSE YEAR(AcDate)
END AS CHAR(4)) + '-' + RIGHT('0' + CAST(dbo.F_ISO_WEEK_OF_YEAR(AcDate) AS VARCHAR(2)), 2)