In working towards the egalitarian ideals of the
revolution IBM Blueview Manifesto, I am proud to present our very first community spotlight interview with IBM Cognos level 99 superwizard Paul Mendelson. Better know as “Cognos Paul”, you’ve likely found success with one of solutions featured on his widely respected blog of the same name – or you’ve unknowingly done so via the many IBM tech notes to which he is a primary contributor.
Paul recently relocated to the United States from Israel and is currently consulting through IBM partner PMsquare. Below is a lightly edited version of an email exchange between Paul and I in April, 2015.
How did you first get involved with Cognos?
This is actually a bit embarrassing. My start with Cognos was a result of laziness and an extraordinary amount of luck. Right out of high school (around 2001) I started a company to build a handheld device to play computer games. Arrow pad, a bunch of extra buttons, and emulators for every OS known to man. It was linux based and the first prototype actually had a regular PC power supply. My partners and I spent all of our money developing it and we managed to get Samsung’s attention. I flew to Seoul twice, and actually shook hands with the CEO (at the time) K T Lee. We were assigned a project manager, and we flew back home to celebrate, and proceeded to hear nothing for the next two months. Apparently they had a management shuffle and my project was deposited directly in the bit-bucket.
Completely broke, and my only skills being having great ideas and playing computer games, I started working in a call center. It turns out that I am really bad at customer service, and even worse at sales. So they pulled me off the phones and stuck me on data entry, transforming Excel sheets into the correct format. It also turns out that, at the time, I was incredibly lazy. So I wrote macros to do everything for me, and played Freecell. When they found out I was terrified they were going to fire me. Instead they put me in the company reporting team. I started building reports in Excel. When it got to the point where running the VBA took a few minutes, I learned Access. Eventually I was given access to a tool called Hyperion Brio, which I hope to never see again. Eventually we got this brand new state of the art BI tool called Cognos 8.2. Compared to Brio, it was incredible. Imagine the difference from 8.2 to 10.2.2, but even more.
Eventually the call center folded, and I got hired as a senior consultant by a Cognos consulting firm.
Can you describe your career a bit? What sort of employers/projects have you worked with?
I’ve worked with consulting firms most of my career. Even when I was in the call center, it was all project based. Since I started working with Cognos, I’ve moved from project to project, and I’m very glad of it. The full list of companies I’ve worked for would essentially a list of companies in Israel that work with Cognos. A few banks, several insurance companies, several government and military offices, a major pharmaceuticals company, a port, a car service/security company, a couple telecoms companies, and a few others that I’m forgetting about. Each company had their own quirks and difficult requirements. One of them needed reports based on a database had a constant flow ~200k rows being inserted every second. Another wanted Cognos completely rebranded. I actually wrote my first article, How to add icons to portal tabs in IBM Cognos Connection based on my work there.
I really can’t emphasize enough how important I feel it is to move from one project to another. Every place I’ve worked at has presented problems that I’ve never considered, or has exposed me to technology that I’ve never had to work with.
What are the biggest changes you’ve seen in BI?
There has been a great deal of iterative, linear changes. Databases get faster, and have more capabilities. BI tools are becoming easier to use, and data is becoming easier to consume (better visualizations, more ways to access the reports, faster run times). IBM and all of the major BI firms are really pushing the self service capabilities, but I’m skeptical. I’ve never gone wrong betting on the laziness of users, and I still expect them to prefer waiting for an IT team to build the reports over trying to do it themselves.The biggest change for me is the introduction of Watson. There has always been a gap between historic reporting (where are the bottle necks in the current production line?), predictive analytics (where is the next point of failure in my production line?), and actionable insights (shut down machine A and get a maintenance team on it before the gears fall off). SPSS is a great tool, but schlubs like me don’t have a chance with it.
What skills do you find most useful and what would you recommend a Cognos developer learn to take their game to the next level?
“Cognos developer” covers such a wide range of tools that it really depends on the direction they want to go in.In terms of data modelling, it’s incredibly important to learn about data warehousing. Read all of Kimball’s books. Read all of Inmon’s. Cognos works best with Kimball’s dimensional design, but you can massage it to work with almost anything. Learn how your database of choices processes queries. Learn as much about DBAing as your local DBA, and make sure you know what kind of candy he likes for when you need some help. Framework Manager and the SQL engine both have their own set of quirks. Framework is actually one of the products where training truly is a must. It’s also important to understand the difference between a relational model, DMR, and Dynamic Cubes, and why/when to use them.
Can you share some of the most avoidable Cognos mistakes you see clients make?
The worst mistake I see is when a client moves from one BI tool to another, and try to make the reports look and behave identical. When people start fighting the tool instead of working with the strengths, the project is doomed to failure. If you’re working hard, then something is wrong. Another problem I see is trying to do everything at once. I’ve been in a few projects were the clients started the report development at the same time as the data model. Hair-loss is almost guaranteed if you’re building a report on a model that’s constantly changing.
Focusing more on Cognos, generally problems occur when people try to use all of the latest capabilities and features. Dynamic Cubes work very well with certain types of data, specifically long and narrow, but when you have only a few million rows DMR should be fine. Analyze the needs and use the appropriate tools in the appropriate manner. While Framework Manager technically allows you to mix DMR and a relational model, it also allows you to build a crossjoin between two multi-billion row tables. Just because you can doesn’t mean you should.
The BI industry is changing rapidly with the move to Cloud, new vendors like Tableau, etc. How would you like IBM to respond?
Cognos does have a cloud offering, and some companies are beginning to open up to the idea. The big security concerns are being resolved.
One of the bigger draws of the smaller BI tools is the advanced visualization techniques. Back in 2009 I was given a demo of SAS JMP and was blown away by the interactive visualization tools. The presenter took data that I provided, and quickly built an interactive dashboard including a 3 dimensional scatter plot. He tied a list to it and clicking on a row highlighted the associated dots in the plot. IBM is getting close with Cognos Insight, but they still need a lot of work. RAVE could be the answer to it, especially if they can find a way to get the interactive graphs in live reports.
Finally, do you have any general advice for people starting a career who may be interested in pursuing BI?
The most important thing for any career is to enjoy it. It pays well eventually, but you have to have a true passion for it to get to that point. A vast amount of my free time consists of reading books or experimenting with Cognos. Most of the BI firms have free or demo versions available. Don’t be afraid to ask for help, but try to solve problems. There’s an indescribable rush when you solve a problem you’ve been working on. Go looking for problems. Trying to solve a problem someone runs into can often give insight into approaches you’ve never considered.
[…] impressive collection of talent: PMsquare recently hired my A-number-one biggest Cognos crush, Paul Mendelson. The opportunity to work with him was too good to pass up, and the fact that PMsquare landed him […]