New MDX Divide() Function
User Rating: / 1
PoorBest 
Written by Chris Webb   
Saturday, 27 July 2013 17:10

Reposted from Chris Webb's blog with the author's permission.

A few weeks ago I saw Rob Collie's blog post about the DAX Divide() function, and I was a bit surprised that I hadn't seen it before. Then, yesterday, I found that the same function has appeared in MDX in SSAS 2012 Multidimensional. Here's the entry in BOL:
http://msdn.microsoft.com/en-us/library/jj873944.aspx

The syntax is:
Divide (<numerator>, <denominator> [,<alternateresult>])

Numerator and Denominator are self-explanatory; alternateresult is the constant value you want to return in case of division by zero, and if it is not specified a null is returned.

It turns out that it was added after RTM of SSAS 2012, and officially first appeared in SSAS 2012 SP1. This is the first new MDX function since. what, 2005? 2008? Five long years at least.

[A loud bump is heard as Chris falls off his chair in amazement]

This is important because there have been requests for a safe divide function in MDX for a long time. I posted a request on Connect, for example, and Darren Gosbell blogged about it here. It would have been nice if someone had mentioned to me that this feature had been added.

Why should you care about this? Well, anyone with any experience of MDX will know that you need to trap for division-by-zero and division-by-null when writing calculations. Consider the following query in Adventure Works:

with
member measures.[France Sales] as
([Measures].[Internet Sales Amount],
[Customer].[Country].&[France])
member measures.[US Sales] as
([Measures].[Internet Sales Amount],
[Customer].[Country].&[United States])

member measures.[France as a % of US] as
measures.[France Sales]/measures.[US Sales]
, format_string=’0.00%’

select
{measures.[France Sales],
measures.[US Sales],
measures.[France as a % of US]}
on 0,
[Date].[Date].[Date].members
on 1
from [Adventure Works]

The measure I've highlighted divides two other measures, and returns the value 1.#INF (infinity) when the measure [US Sales] is null:

image

1.#INF is not something you want to show to your end users. Furthermore, Mosha showed us all a long time ago that when dividing we should always use the pattern
iif(measures.x=0, null, measures.y/measures.x)
to get the best performance.

It now looks like this pattern has been superseded by the Divide() function. Here's the measure in bold from the query above rewritten to use it:

member measures.[France as a % of US] as
divide(measures.[France Sales],measures.[US Sales])
, format_string=’0.00%’

From my limited testing on Adventure Works performance seems to be the same as with the iif() pattern, but I have heard that in other scenarios it may perform better. So I would recommend you try testing it on your cube, and use it in all your MDX from now on.


chris-webb

Chris has been working with Microsoft BI tools since he started using beta 3 of OLAP Services back in the late 90s. Since then he has worked with Analysis Services in a number of roles (including three years spent with Microsoft Consulting Services) and he is now an independent consultant specialising in complex MDX, Analysis Services cube design and Analysis Services query performance problems. His company website can be found at http://www.crossjoin.co.uk and his blog can be found at http://cwebbbi.wordpress.com/ .


 

 

XL Cubed