Profilo di MichaelMichael's BI SpaceFotoBlogElenchiAltro Strumenti Guida

Blog


30 settembre

Algorithmic Depth

I've always wondered why there is so little focus on the Intelligence aspect of BI - something that just confirms my belief that BI is more commonly used to describe base reporting.
 
For the basics - Peteris Krumins’ takes time out to share MIT lectures: http://www.catonmat.net/blog/mit-introduction-to-algorithms-part-seven/
 
And there are q few comparative documents around - note: Undetermined validity http://www.telefonica.net/web2/todobi/Dic07/ComparingWekaAndR.pdf
 
Generic method to use for product evaluation: Place into a spreadsheet with the Yes / No  / Other options to rank the features. 
 
GENERAL
Add-ons
Libraries
Price
Required Software
Supported platform   
APIs and system interfaces
APIs   
Interface
Language support for pipes, sockets, and files
Reports   
Data and graphics formats
Application data
ASCII
Database
Export file type
Financial data   
Import files   
Native database   
ODBC interface   
Spreadsheets 
  
DEVELOPMENT ENVIRONMENT
Debugger   
Framework   
Libraries   
Programming Language   
Project, file and task management   
User-defined functions, objects, classes, methods and libraries 
  
GRAPHICAL FUNCTIONS
3-D wireframe charts   
Bar charts   
Box plots   
Control over colors, lines, axes, annotations and layout   
Dot chart   
Histogram   
Interactive charts   
Pie charts   
Scatterplot   
Time series charts 
  
GRAPHICAL USER INTERFACE
Create custom toolbars, menus and dialogs   
Cut-and-paste to Word, PowerPoint and Excel   
Database import and export dialogs   
Drag and Drop   
File import and export dialogs   
Integrated Excel spreadsheets   
Interactive command-line with history recall   
Manage objects with Object Explorer   
Multiple data and graphics windows   
PowerPoint Wizard: quickly create slides from charts   
 
SCALABLE PIPELINE ARCHITECTURE
Data types for out-of-memory vectors, data frames, and time series   
Scalable algorithms    Scalable model estimation   

STATISTICAL & NUMERICAL TECHNIQUES
Analysis of Variance   
Cluster Analysis   
Correlated Data Analysis   
Correlation    Crosstabulations   
Date, Time, and Calendar Data   
Durbin-Watson   
Goodness-of-Fit   
Mathematical Computations   
Maximum Likelihood   
Missing Data   
Multivariate Analysis   
Neural networks   
Nonlinear Regression   
Nonparametric Regression   
Power and Sample Size   
Probabilities, quantiles, densities and random number generation   
Proportions   
Quality Control   
Rank tests   
Regression   
Resampling   
Robust Statistics   
Summary statistics   
Survival Analysis   
t-test   
Time Series Analysis   
Tree Models   
Wilcoxon   
SUPPORT On-line
29 settembre

Head in the BI Clouds

http://en.wikipedia.org/wiki/Cloud_computing are all the rage these days - and it will impact BI - here's a list of the some of the top 20 playa's
 
Moving on back up to BI - there appears to be a never ending range of solutions out of the Gartner spotlight:
 
And always turn around and look back at your solution stacks from a management perspective:
 
28 settembre

MS BI

Many think that I am an unthinking antagonist for all of Bill's (read Microsoft) products.
Not so - when it comes to BI tools I simply think that the generic toolsets, that are database independant, have a slightly different focus in versatility.
Sure, the counter argument is that these tools will optimally leverage the database environment for maximum data usage.
Until such time as an organisation has matured in it's use of toolsets and technical, analytical and business interactions, it might be wise to keep some of the technology out of the inner core IT department.
 
It will be interesting to see how MS reacts to the next generation of tools that will piggyback on their products:
 
Gui meets BI, but misses something http://www.lyzasoft.com/features.php 
25 settembre

BI & DWH Appliances or SaaS ?

The Business Intelligence and Data Warehouse appliance industry has been hovering around for a while, but I've only managed to get physical access to http://www.cognos.com/products/now/ 
Others are following the route of offering the entire package as a service instead http://www.kognitio.com/services/businessintelligence/daas.php 
 
 
Goes to show how anything can be packaged - and it will be interesting to see if the industry matures enough for more people to rather opt for the prebuilt or off site solution as opposed to inhouse and self built.
We've seen this cycle constantly in IT and most orgs tend to reach that point of just not wanting to hear about any new issues relating to upgrades, incompatiblity, resource, performance, lincense mixes etc  
 
Others are calling the appliances glorified data storage units - so lets see where this all goes. 
 
And don't forget to keep watching the Open Source space http://www.sun.com/solutions/landing/infrastructure/dwa/ 
24 settembre

BusinessObjects Migration - XIr2 or XI 3.0

 
The underlying issue is that XIr2 and XI 3.0 are NOT version upgrades from BO 6.x.
This complicates the issue of which version to migrate to, as the one is an interim consolidated version, based on previous Crystal architecture, whilst the latest is the reworked consolidated version that will be used going forward - that in turn is UNPROVEN.

And - it's not entirely correct that SP1 will offer the secure base most refer to as XI 3.1 will be a FIRST release of the .Net version and IIS.
So please consider that my UNPROVEN statement is wholistic in a dual sense.

There are a number of guiding questions one could pose to the organisation to help determine which version one should migrate to, but the bottom line is - do you have the resources, architecture and time to do an effective migration using a phased approach, or is your migration window and access to resources and architecture limited - to a big bang, one hop now?

Governing this is the fact that you WILL be required to get to XI 3.x, at some stage - so the question is, do you use XIr2 as an interim hop to get there?

I follow the debates in a variety of places - and receive a number of emails on this subject, and it's always a case of, which one to migrate to - instead of questioning if a two step approach should be considered.

Most don't want to hear about a possible 6 month to 2 year journey to get there - be that from a corporate or technical perspective.

The determining factor will always be - the organisations dependancy on BO and their relationship with you !!!!!!!!!!

I do not question the success stories in here - and already have a set XI.30 servers up and running - so, as much as I am convinced that XI 3.x is the way to go - strategically it might be too risky for some.

The ideal would be - and has been mentioned before - that if you start now with XI 3.0, you should be able to time this nicely with the XI 3.1 release - which is what I'm hoping to convince my current client to do.

BUT

I would be hesitant to stake everything on this - and am the first to state that there IS a risk so the parallel option is unquestionable:
I.e. do not hope to free up your current servers for this process.

I have an unholy mix of 6.5 and stalled XIr2 installations that I am looking at consolidating into XI 3.0 - so the issues are more complex in that I have scattered solutions that are not in synch.

I am not convinced that the XI 3.0 release is optimal from an architecture perspective - and am waiting patiently to see some benchmarks - from the vendor and from some of the trials that I am performing.

It is a slick, nice effective portal - the architecture is better - but still needs some tweaking - so that poses another set of questions:
Will you effectively be able to hand hold users faced with a new environment
At the same time resolving some security issues (there will be a settling in period no matter how effective you establish your CMS environment)
And whilst possibly dealing with the odd backend issue ?

It's also not that simple to just segregate the issue between 1 / 2 server installations, and complex 24/7 server clusters either.
The latter typically involves more people in the decision making process and typically forms part of an architecture governance program whilst the former could be a single server delivering event tracking that is core to the running of the business.

Add the fact that the learning curves are STEEP on ALL sides: Security, Usage, Admin, etc

So the debate then involves:
Does it make sense to use XIr2 as an interim learning mechanism as there are alot of similarites and lessons that could be taken going forward?
Or would the duality of this be a time waster in that any effort applied to this process will have to, in some respects, be duplicated?

If I knew I was going to start posting this, I would have been better off creating a set of relative questions to be asked to help determine this - regardless of your installation size.

I.e As it is with all things DWH / BI related - it's the softer issues that have to be discussed up front with the client.
A frank honest discussion with the business as to the pro's and con's - as well as the risks and benefits - and then the respective time frames and learning curves should form the basis of the discussion.

Companies don't like being surprised after the fact - and if you involve them up front - all the way through - you can make a success of this - even if it turns out that some of the platform is sub-optimal for a while whilst the vendor resolves things.
That is if you plan for it ... and keep everyone in the loop.

Note that there are external influences like local vs external resources - access to architecture and licensing etc etc that I have avoided for now - maybe I should create list - but I stillowe BOBJ a development life cycle plan icon_redface.gif so am avoiding any further promises icon_rolleyes.gif

Also add to the above the fact that you have a bunch of previous IIS admins that face a Tomcat curve ....
23 settembre

XI Tech dragging admins to Jakarta

I was ecstatic  that the XI releases from BusinessObjects focussed on Tomcat as opposed to IIS - but - that has a bunch of admins that are pretty clueless as to the Jakarta offerings.
To complicate matters further, XI is releasing the Enterprise BI product onto Tomcat as a Web AND Application server - basically replicating the IIS dual functionality, and when faced with a requirement to split this functionality, many opt for throwing IIS in front of Tomcat as opposed to Apache.
I migrated an enire site off IIS back in the old BO version 6 days - and have never looked back - I do admit that it was quite a change and bought a manual that helped me along, but many online sites were either too much on an idiot level, or waay too complex to get a general understanding as to the possiblities of Apache / Tomcat.
Here's a pretty good generic overview http://www.easywayserver.com/
 
Microsoft showboating again to keep in the BI news http://www.calumo.com/blog/
 
 
22 settembre

Off the Beaten BI paths

Ever notice how all the 'prime' technology sites tend to regurgitate the core usual suspects with respect to news on the BI front? 
I always make an effort to try and see if there is anything else happening:
 
DWH & BI terms and abbreviations: http://www.sdgcomputing.com/glossary.htm
 
And a nice overview of the SOA/BI edge: http://www.infoq.com/articles/BI-and-SOA
18 settembre

BI Modelling

Once in a while you encounter a blog that has some really valuable content:
 
..... aka Sassy data Chick - I applaud thee !!
 
Anyone struggling with 32 bit applications on 64 bit architecture - take all advice with a pinch of ..... cosmic dust
 
17 settembre

InfoViz - Information Visualization

http://en.wikipedia.org/wiki/Visualization_(graphic) is critical in that the visual appeal of something on a primary sense, occurs faster than the rationalisation of a logical determination process.
Ever since humanity discovered how to create purple dye, the quest for colour has driven a number of industries - and the frontier for this has shifted to the virtual plane.
 
And yes - they invariable find their way over to BI - so don't for a minute think that the BI vendors ared driving this concept - it's as old as a sabre tooth fossil and probably a mutant of scientific research institutes and game developers .....
 
 
 
16 settembre

Intellect - a commodity

I once 'worked' with a professor in Maths and Stats from Zurich, who lost us totally after only about 2 hours when trying to explain how database management systems REALLY work. This was when most core databases were mainframe based (DB2 - MVS and VM), and I oft wonder what it would be like to encounter him again after more than a decade of rampant technological change.
In the RDBMS world, the database and DBA toolsets have got smarter and more complex - but do most DBA's actually know and understand the underlying 'engine'?
Once in a while it's worth looking into the labs where the true thinkers exist - if only to humble ones self concept of knowledge:
 
Whilst the majority of online browsers have a misconception that anything can be had via online browsing aka google and chrome, there are still places that mainatin the basic principle that thought and conceptualisation hold an inherent quantifiable value.
 
 
So - just when you thought you reached http://www.endoftheinternet.com/ and know where and how to retrieve anything - go back - do not past go - and haul out your paypal or credit card for the data vaults of information that have been quantified as being worthy of an intrinsic monetary value.
Catch 22 really as it's all topical - they all have shelf lives that can expire today or somehere in the next ice age.  
15 settembre

Intelligence - BI-OI-LI-CI-?

Regardless of all the 'BI' headlines, there be many forms of http://en.wikipedia.org/wiki/Intelligence for corporations.
No intelligence set is complete without including all information aspects that influence or govern a thought, event, process or entity.
And some organisations are well aware of this http://www.ventanaresearch.com/research/overview.aspx?id=1179
LI - http://en.wikipedia.org/wiki/Location_intelligence - and I had to roll my eyes at the fact that www.locationintelligence.net/ timed out ....
et set ra ....
 
Be that as it may - automation and autonomy is driving the pace and this 'consolidation' must certainly look like a holy grail from way up above the coroporate food chain
 
And then again - best watch this lot http://www.intelligenceonline.com/ or join them https://www.mi5careers.gov.uk/job.aspx?jobid=147 
 
I say - call it what you will - my resume is open ended enough to convey the fact that I can process bytes into bitmaps - on a variety of platforms - using a mix of technologies - to hopefully service any realistic demand.
14 settembre

BI Assessment

TDWI is quite consitant in it's assessment tool release - low level and requires input etc http://www.tdwi.org/display.aspx?id=8500
Makes me think of their precursor to this - an excel spreadsheet that was related to Federated DWH - something you don't hear about anymore. 
 
As to assessments - there be many - from strategic to technical.
 
But the basic principles don't change, and you need to ensure that your assessment is complete in ALL respects - this link is dated,  but is quite holistic in it's intention - and the embedded truths still stand http://www.tdan.com/view-articles/4873/
11 settembre

New BO Global Community

Physical Communities evolve all the time - I've seen user groups come and go depending on the people and organisations involved - and many have died due to the virtual realms springing up everywhere.
Dave Rathburn is involved in a new BO cummunity endeavor - which is the only reason I would consider watching this space
 http://www.gogbn.org/ as I've seen the global user communities die during the recent mergers and takeovers.
 
 
  
09 settembre

IDE & BI

http://www.eclipse.org/ has always been a favourite tool of mine and we're seeing more an more BI vendors use this IDE to expand and integrate BI solutions.
But it's always good to see people try to provide good objective critique when comparing toolsets - we are a very preferencial species and are more vendor loyal than we care to admit http://dev.eclipse.org/newslists/news.eclipse.birt/msg12522.html
 
And down south a Sybase - Yellowfin - eVolution partnering http://www.theserverside.com/discussions/thread.tss?thread_id=50376
 
So yeah, TDWI and Gartner will have you think that the "leaders" are where the BI front is, but the smaller units are pushing forward regardless - who knows where one territory ends and the next begins .... 
06 settembre

BI - Write Back Mountain

Data Transformation spent years scaling the GIGO mountain with bad transactional / operational data quality issues - hopefully the SOA / Web Services crew are taking this into consideration with respect to write back capablities!
Excel has always had the one feature that NO BI / Reporting tool set has been able to offer the end user - immediate data write back and update capability - but hopefully closed loop BI is finally upon us. 
SOA and web services are already making deep inroads into this oft ignored territory and sure, the uber techies and business entities are quite excited at the prospect of not having to develop or integrate backend applications that sometimes span a decade of difference in technologies, but I hope the data custodians have learnt the importance of Meta Data and Quality Control ....
 
TAWS - Take Action Web Services from BO has been upgraded http://labs.businessobjects.com/take_action/default.asp
With all the acronyms and technologies available it's quite easy to confuse WebFocus with WebSphere - but there is a difference http://www.e-omnisys.com/omnpg_files/ClosedLoopBI.pdf
                                                                                           ......................................
Interesting concept - lacking in real data - but you never know http://kpilibrary.com/
 
I've somehow managed to avoid the dark realm of pure java application development - hopefully sites like this makes it all worthwhile http://www.manageability.org/blog/stuff/open-source-java-business-intelligence
03 settembre

The '6' pack

Continuing on with the dashboard concept being the washboard or 6 pack abs part of a BI anatomy i.e. looks cool - attracts attention - very functional pivotal part of the extended extremeties etc - the core issue is one of proprietary vs third party.
Many of the larger BI vendors slowly worked their way up to the dashboard and scorecard capability stage - most by purchasing smaller vendors that had already been down this road. Those that were not bought out, made use of this lag by specialising in generic type solutions - typically including in excel and the core RDBMS flavors - and they have not stopped offering solutions that wrap around the big gun's solutions either.
Others focus on solutions  
 
The question is: Knowing that the big vendors are banking on the 'account consolidation' premise re corporations long term strategy - how viable are these more dedicated solutions?   
02 settembre

Screen Abs

As it is with the 'hot' topic of washboard abs - elusive in nature - but silently desired - so it is with Dashboards and Scorecards and it looks like non topical vendors are quietly getting things done - http://www.microstrategy.com/digital-dashboard/demos.asp 
 
Or is it just reworking and relaunching existing technology at a more opportune time? The marketing game is a devious one - and relies on 'push' methods - so one cannot fault the vendors for ensuring that they stay topical, and select quiet windows inbetween 'big' release notices. Makes me wonder how many tech orientated analysts have been waiting for the MS SQL 2008 headlines to fade .....    
 
Generic site with example of Enterprise Dashboards - the only drawback being that there are no examples based on the latest software versions.
 
01 settembre

Time for Tech

Most used command for remoting onto servers:  mstsc /v:servername /f -console
It doesn't always work - but in most instances it will help grab a session.
 
Automating XI backups:
 
REM       ------------------ XI Backup --------------------
 
REM Stop Core Services - ensures no post system shutdown sync problems with CMS and FRS
net stop "Central Management Server"
net stop "Input File Repository Server"
net stop "Output File Repository Server"

Rem Date variable for folder
set dateString=%date:~10,4%%date:~4,2%%date:~7,2%
mkdir c:\BOEXIBackup\%dateString%\
 
REM Below:  /E = trusted login - can replace with /U and /P (username and password)
osql /E /Q "begin declare @CMSBackupFile varchar(255) set @CMSBackupFile = 'C:\BOEXIBackup\' + CONVERT(VARCHAR(20),GETDATE(),112) + '\CMS' + CONVERT(VARCHAR(20),GETDATE(),112) + '.bak' backup database {CMS_RDBMS} to disk = @CMSBackupFile end"
 
REM Audit backup
osql /E /Q "begin declare @AuditBackupFile varchar(255) set @AuditBackupFile = 'C:\BOEXIBackup\' + CONVERT(VARCHAR(20),GETDATE(),112) + '\Audit' + CONVERT(VARCHAR(20),GETDATE(),112) + '.bak' backup database {AUDIT_RDBMS} to disk = @AuditBackupFile end"
 
REM Copy FRS (Input and Output) to the backup folder
xcopy "C:\Program Files\Business Objects\BusinessObjects Enterprise 11.5\FileStore\" c:\BOEXIRBackup\%dateString%\
 
REM Start Core Services
net start "Central Management Server"
net start "Input File Repository Server"
net start "Output File Repository Server"