More servicesWindows Live
HomeHotmailSpacesOneCare
 
MSN
Sign in
 
 
Spaces home  The BizSharpie Performan...PhotosProfileFriendsBlog Tools Explore the Spaces community

Blog

October 01

PerformancePoint Server Overview Webcast

A webcast overview of PerformancePoint showing all the main components of PPS has been posted here:
 
TechNet Webcast: Performance Management with PerformancePoint Server 2007 (Level 200) 
 

Webcast Overview

In this webcast, we explore how Microsoft Office PerformancePoint Server 2007 can help corporations improve their business performance process by integrating monitoring, analysis, and planning capabilities into one application. Join this session to learn about pains customers commonly experience with performance management, and how PerformancePoint Server 2007 enables businesses to deliver performance management capabilities broadly by providing an application that is easy to use and less costly to deploy than traditional solutions.

Presenters:
Bruno Aziza, Senior Product Manager, Microsoft Corporation
Alyson Powell Erwin, Program Manager, Microsoft Corporation
Patrick Baumgartner, Program Manager, Microsoft Corporation

September 28

PerformancePoint New York Launch Event

The New York launch event for PerformancePoint was big hit. The demos went well. Chelsea Piers was a great venue - I snapped a few pictures during the day and they are linked...
 
The webcast available here. And here is some press and analyst coverage.

We have more events to come... The Microsoft Virtual Business Intelligence Conference on October 4th (register here), or the London launch event on October 16th (check it out here).

Keep up to date by visiting the PerformancePoint Server 2007 site here. And, of course, if you want to download the RTM software, visit Microsoft TechNet here.
See you in London!
 
September 17

Launch Event this Week...! And another case study has been posted...

Another case study has been posted. This one is from Energizer...
 

Energizer:

Energizer Taps Into the Power of Microsoft Business Intelligence with Microsoft Office PerformancePoint Server 2007

http://www.microsoft.com/casestudies/casestudy.aspx?casestudyid=4000000646

And the biggest news, of course, is that we've got the first PPS launch event this Thursday in NY. I'm planning to take lots of photos and will post them up here...

September 04

Case Study for PerformancePoint Server

A new case study for PerformancePoint Server has been posted.
 
Oticon, a hearing aid manufacturer based in Denmark, is using PPS for financial consolidation and planning, budgeting, and forecasting. They are working with Microsoft partner Solver Inc. on the implementation.
 
Oticon

Oticon Hears The Value of Performance Management with Microsoft Office PerformancePoint Server 2007 

http://www.microsoft.com/casestudies/casestudy.aspx?casestudyid=4000000624

August 07

General “tips” for investigating and improving performance in the PPS Add-In for Excel (in no particular order…)

Based on the steps we take when investigating performance issues with the PPS Add-In for Excel, I thought it may be helpful to post some of the things we check for. In no particular order, here is what we walk through when trying to improve performance for a form or report in the PPS Add-In for Excel. 

1. Compare the results of running queries in the PPS Add-In for Excel vs. directly against Analysis Services

a.    Use the “View MDX” option from the PPS menu/ribbon to copy the MDX for a matrix. Use SQL Management Studio or other tool to run this MDX directly against Analysis Services*. If the time is similar then investigate optimizing or changing the way calculations are defined in the model. *Note: You will need to adjust the MDX query to return results on 2 axis to run it in SQL Management Studio by moving the ‘on pages’ portion of the query to the where clause or other approach to making the query result render-able by SQL Management Studio.

 2. (Excel 2007 Only) Set the matrix style to none

a.   The matrix styles feature adds additional processing time to format the matrix. Setting the matrix style to none will speed up performance by removing this step.

3.   3. Check if  the time the add-in is spending “detecting writeable regions” is a bottle neck. If so separate the writeable areas and read only areas in a matrix into separate matrixes.

a.    If a matrix used in an assignment has “Allow Data Entry” set to true, then the Add-In will do additional processing to determine what cells are writeable and highlight them. If setting the “Allow Data Entry” property to false speeds up performance significantly you may consider separating the potentially writeable areas of a matrix and the read-only areas of a matrix into separate matrixes and setting the “Allow Data Entry” property to false on the read only matrix.

4.   4.  performance “gets slower” over time while entering data in an assignment it may be because the “What-If Change-List” is growing large and thus taking more time to calculate.

a.    Using the "View change list" option you can view the current contents of the assignment change-list. If it appears your performance problem is occurring as this list grows, consider doing the following:

                                          i.    Break the data entry matrix in to multiple smaller matrixes. The “What-If” query will only apply over the matrixes where changes are being applied, so data unaffected by the what-if changes will return faster when in separate queries.

                                        ii.    Consider instructing users to do use the “Submit Draft” option to submit changes to the server that the user is done changing, use the “Clear Changes” option to clear the change-list, and then re-download the assignment cache. This could help keep the change-list size smaller and lower the amount of time for recalculations

5.   5. In general, the first time a query is run in a report or assignment it is usually is slower. Additionally if it is the first query executed against the Analysis Services server it will also be slower. Check and see if performance improves on subsequent refreshes.

6. Check the size of the local cube being downloaded in an assignment where the "Automatically cache assignments" option is turned on.

a.    When an assignment is cached locally a local cube is created to store data and perform recalculations. If this cube is large the client machine may have difficulty downloading and querying against it.

b.    To check if this is an issue clear the offline cache and disable the automatic caching feature to see if “online” performance is significantly better. If it is investigate making the size of the local cube that will be created by:

                                          i.    Reducing the read permission scope for users

                                        ii.    Disabling the “offline” functionality for the model in BizModeler which will disable the automatic caching feature for all assignments that use the model.

                                       iii.    Using the “manually define sub-cube” option to define a more specific sub-cube to download and cache locally.

 7.    Try to use calculated members over Design-Time formulas whenever possible

a.    Rendering and processing of design time formulas by the add-in is a common performance bottle neck.  

May 02

We have a question - Why am I not seeing aggregated data in this form?

This blog has been alive for a couple of months now, and we finally have a question to answer (If you have questions please send them and we’ll track down questions for them too... )

Question:
(The scenario is Activity Based Costing for doing localization of a product)

  • I am a modeler with full data permissions and I have a form assignment on a Generic model with some data submitted for it
  • The model has a few SQL rules to populate data, and a few MDX rules that calculate various slices of the cube not used by the form assignment
  • It uses the following pre-defined dimensions: Scenario, Time, Currency, TimeDataView

It is linked to two generic assumption models containing proper subsets of its dimensionality.

Exhibit A has a preview of the assignment working properly. (See image below)

After submitting my data, I decide I’d like to see the total across all my submissions, so I multi-filter on my Spec-ID dimension and include items 1 and 4.  Even though Spec ID 4 contains no data, I want to look at the two together. 

When I do, I get what is shown in Exhibit B. (See image below)

There is no data :(

“So…”, I ask myself, “what happens if I multi-filter on the name property instead of the actual members?“ 

Using the “Add a dimension property filter” feature,  I remove [Spec ID].[Member Set] from my filters, and add [Spec ID].[Name] and I get Exhibt C.  (See image below)

The ‘View MDX’ feature instantly becomes my new best friend and helps explain the differences between Exhibit B and Exhibit C, but I still don’t know where the bug is.

What is going on?

(Thanks to Luming and Peter on the PPS Excel Add-In team for helping track down the answer…)

First, for what is happening in the different “Exhibits.”

In the PPS Add-In for Excel when filtering with multiple members selected we use a calculated member to get the total (which is the same as a PivotTable). The drawback of this, as you see, is that if the value cannot be aggregated across the dimension, you will see the Exhibit B behavior. On the other hand, in Exhibit C, filtering on dimension properties uses a subcube, so you still get the totals.

Beyond the self identified work around in “Exhibit C” (which is what makes this such a great question…) how can we fix this for Exhibit B.

To start, sometimes the non-aggregation behavior is what we want. For example, if the dimension in question is “currency”, we wouldn’t want to aggregate the values on the currency dimension. It doesn’t make sense for an aggregation of U.S. Dollars + Japanese Yen. So if you select the filter as “{USD, JPY}”, you would get nothing back.

To change this behavior we need to check the aggregation operator/rule on the questioned dimension. If the measure is semi-additive and cannot aggregate on the questioned dimension, we’ll get non-aggregate behavior. However, if we switch it to additive you should get the aggregation behavior you’re expecting here.

March 25

Microsoft Business Intelligence Conference 2007

Microsoft is holding its first ever worldwide BI conference from May 9-11 in Seattle. If you haven't registered do so now because I understand the event is filling up fast:
 
There will be several break-out sessions covering PerformancePoint and the keynote speakers include Steve Ballmer, Jeff Raikes, Ted Kummert, Dr. Robert Kaplan.
March 07

How to open a PerformancePoint assignment from a hyperlink (e.g. I want to send an email with a link that automatically opens an assignment)

I received a question a few days back asking if it was possible in PerformancePoint to send out links to assignments and just saw something similar on one of our DLs so wanted to give an explanation of how do this.

First, where this would be useful is if you wanted to put together a portal page with a web part containing links to the current user’s data entry assignments, open PPS assignments from another application, send emails to contributors with these links, or otherwise make it easier for users to quickly open Excel and get working to submit their numbers.  

To facilitate all of this the PerformancePoint Add-In for Excel ships with a protocol handler that, after registered on a client machine, allows you to specify a hyperlink such as “PerformancePoint:http:\\PPSServer:46787&Application=AdventureWorks&Assignment=1027” which when clicked will do the work of opening Excel and opening the assignment that is specified.

When you install the PPS Add-In for Excel the protocol handler is dropped into your install folder with the name “PPSProtocolHandler.exe”

To register the protocol handler on a client machine, from the command line navigate to your install folder and run “PPSProtocolHandler.exe –REGISTER:M” (we don’t register it by default today as we’re expecting IT to want to control its deployment)

As you can probably intuit from the example string above the format for the protocol handler is pretty straight forward:

PerformancePoint:http:\\<Server>:<Port>&Application=<Application Label>&Assignment=<Assignment ID>

Where:

<Server>:<Port> == the server and port number of an applicable front end server in you deployment

<Application Label> == the label of the application where the assignment you want to open resides

<Assignment ID> == the ID of the assignment to be opened.

After registering the protocol handler on a client machine and creating a hyperlink as indicated above clicking it will launch Excel and open the indicated assignment.  Put this together with a little bit of code to fetch the ID’s of the current users assignments (a follow up topic for another post) and you can start sprinkling in links to PPS Assignments in the context of your existing portal or wherever users want to manage their work from.

March 06

Getting Started

I've been promising to start a PerformancePoint blog for almost a year now - so here we go...

As the title of the blog would suggest, the content will focus on the PerformancePoint Planning (formerly known as Biz#) component of PPS. I'm hoping to call on some of my teammates on the PPS team to add some blog posts here as well so we'll also be able to span a wide variety of topics relating Performance Management and PerformancePoint. If you have any requests let me know.