The BI Guy
-
Regarding BI in major corporates.
The following anecdote is not me bragging, as you will see when I get to the "punchline".
I implemented a global OLAP intranet-based reporting tool to handle Assets Under Management for a big Asset Management company. This is based on an Office Web Component pivot against MS SQL Server 2000 Analysis Services, and it handles 17 dimensions, pulling data out of the multiple systems around the world, etc., with multiple currencies, etc. So it is fast, cheap and the business like it.
I also implemented a very large IIS/ASP intranet reporting system that covered most areas of the business ( trading, portfolio management, ops, etc. ) using Office Web component pivots and charts against tabular data sources, so the users had local OLAP. Everybody in all the offices uses this system, which runs locally in each main office for the past 5 years
Here comes the "joke" :
About 5 months ago the company made java "the strategic development platform." It is acknowledged by all the developers and the technical architect that there is no way to implement this dynamc pivot and chart functionality in a java web site. However, that does not seem to matter, and no more ASP or ASP.net development is permitted. MS SQL Server can not be used for any new, strategic development.
I left the company ( and the large salary and bonus ) 3 months ago.
BI works, even on a large scale, but, if the IT management do not get it, you are stuffed
Leo
-
Well, I see a lot of people have already alluded to the BI concept, but my personal idea on BI is a tool that allows you to analyze and interpret information about your business. Be it OLAP/Cube based or Dimensional Database based.
As far as OLAP, and datawarehousing, Yes, we have been doing it for several years.
My departement is called "Enterprise Information Management" and our job over the last 3-4 years has been building and maintaining a complete Dimensional datawarehouse backend, and the Cognos BI Tools front end that utilize it. The datamarts themselves are in SQL Server, and use the Kimball (star) dimensional design. (We extensively use an ETL tool, Cognos Decisionstream, to support building the Dimension and Fact tables)
On top of that we have various Cognos PowerPlay "cubes" as well as other reports and tools that use the database directly to perform BI functions. Our users use this data in a BI respect extensively, especially in product development and marketing, as well as accounting.
Our next steps here are in leveraging the information for executive scorecards, and perhaps dabling more into integration and CDI. (Customer Data Integration)
-
( Previously posted to an empty forum )
Regarding BI in major corporates.
The following anecdote is not me bragging, as you will see when I get to the "punchline".
I implemented a global OLAP intranet-based reporting tool to handle Assets Under Management for a big Asset Management company. This is based on an Office Web Component pivot against MS SQL Server 2000 Analysis Services, and it handles 17 dimensions, pulling data out of the multiple systems around the world, etc., with multiple currencies, etc. So it is fast, cheap and the business like it.
I also implemented a very large IIS/ASP intranet reporting system that covered most areas of the business ( trading, portfolio management, ops, etc. ) using Office Web component pivots and charts against tabular data sources, so the users had local OLAP. Everybody in all the offices uses this system, which runs locally in each main office for the past 5 years
Here comes the "joke" :
About 5 months ago the company made java "the strategic development platform." It is acknowledged by all the developers and the technical architect that there is no way to implement this dynamc pivot and chart functionality in a java web site. However, that does not seem to matter, and no more ASP or ASP.net development is permitted. MS SQL Server can not be used for any new, strategic development.
I left the company ( and the large salary and bonus ) 3 months ago.
BI works, even on a large scale, but, if the IT management do not get it, you are stuffed
Leo
-
We have a series of data marts in SQL Server, Analysis Server OLAP cubes and OLTP databases that are used in enhancing and modifying the raw data for business purposes. Front end tools are Reporting Services, ASP, Win Forms, Access and Excel.
So far in the BI world he have a small implementation. I think the wrong approach to BI is to be pidgin holed into a single solution or tool. This is why we are staying with a Microsoft toolset. Users can get the data in whatever format that they are most comfortable with. Development though is definitely more of an art than a science.
-
...and you're implying that "business" and "intelligence" are often highly correlated together, which most of us would probably argue is not the case, unless you're a CxO/SVP padding your stock options or waiting to cash out in your golden parachute.
-
All,
We've had a very successful implementation to about 10,000 users which we put live in 2001 on MSAS and a 3rd party front end (targeted specifically at our business type). It looked like it was going to be a real chore but we made a fairly radical (for us!) move at the beginning which made it much easier. Before the project officially kicked off we separated out responsibility for Information from IT. The MIS team became entirely separated from the IT department and reported directly into the Business.
This meant that we (the MIS Team) made decisions and developed a system designed purely to meet the needs of our Sales / Service force. The part of the team which developed the KPIs and business rules were a mix of operational planners and previous sales staff, the developers were previous IT DB people (myself included). The beauty of OLAP means that we have a reporting tool and an Analysis tool sat on the same cubes and base data (there's 127 different data inputs rolled into the beast, although nowhere near that many when we rolled out in '01!) with the Sales and Service force able to view yesterday's performance or analyse their own or team's particular metrics over a year or so to spot their own personal best practices and peaks. The exec use it at a much higher level of the pyramid to realise project benefits and steer on major policy, channel direction decisions.
We still need to sort out our drill through... I personally believe MSAS missed the boat when it came to drill through in 2000 and I'm not ready to look to see whether 2005 has implemented it any better (anyone tried?). This has left one vital part of transactional reporting looking at the raw FACT table rather than the cube but apart from that... fantastic!
Don't get me wrong, there are dis-advantages to splitting off from our brethren in IT (or are they just, "T" without us?). Over the last 5 years many staff have changed and when we ask them to buy and build us a specific server now there aren't many left who know we are ex-IT ourselves and feel they need to constantly question our requirements. That said... the ease of implementation, the immediate buy-in and use from our internal customers, the ability to quickly adapt without being delayed by "pending" large IT deliverables and the heads-up we get on new business directions (allowing us to code for a new products *before* they go live!) massively outweighs these occasional niggles.
Happy to answer any questions on it. I can understand how many people in IT would be nervous at a break from it but... on a purely personal note... my yearly bonus got a lot bigger when I joined the business side of things - they really appreciate the difference good BI can make!
-
One of the products my company does uses OLAP to provide BI to management at varying structures. Previous versions simply used SQL. All AS does is provide extra performance and architectural advantages.
Still does not help with getting the proper requirements out of the customers though!
-
Hm... Some good points in there - I've been looking at SQL server 2005 Analysis services and frankly I couldn't be bothered. I have actually got a fair way with Qlikview now and it is very promising technology. However, the maths behind it is not very strong, I've found it is better to summarise your data in the background and just hook into the results.
So - my process is SSIS to grab the data - SSIS to transform it so I can relate the tables I am populating together on a one for one basis - Qlikview to analyse the data.
Interesting point about using existing reports - that is exactly what I'm doing - I figured if you're going to give the users the power to analyse data you'd better make sure they understand it. So, I'm taking the existing operational reports and stitching them together.
The hardest part of it all is convincing users to let me know what they're up to and why and how they're doing what they're doing - the company is full of people who think they're doing a great job labouriously putting together huge spreadsheets and they don't need IT for anything (unfortunately a sentiment shared in large part by my MD). Touche whatever idiot slagged off IT departments for wanting to buy kit - must've had a lousy IT manager.
What they don't realise is that if they only allowed me to analyse their problem they'd never have to do it again and the resulting applications would be far more powerful - half my job is trying to guess what kind of pointless spreadsheet they're going to decide to fill in next.
-
You could always try what we did:
Me: "Now you have this great BI system available to you, you can do all your planning in it."
Customer: "Yes, yes we can. Thank you."
Me: "So you don't really need Excel do you?"
Customer: "Well... we.."
Me: "I'm sure you wouldn't want the company to pay twice for the same functionality?"
Customer: "Well of of course not but...!"
Me: "Well done, I think you've just saved us a fortune by scrapping your Excel licences. Let us know if you need any more training on the BI tool."
Customer: "Yes... erm.... thanks?"
Me: "I'll be sure to mention to your department head how much your forward thinking approach has save the company."
Believe it or not... something similar to this worked on 8,000 out of our 10,000 users. We should get about 1,900 of the others once I sort out a Drill-through or Reporting Services bolt on.
Viewing 9 posts - 16 through 23 (of 23 total)
You must be logged in to reply to this topic. Login to reply