What is going on in the BI/DW world?
By Jean-Pierre Dijcks on Jun 23, 2009
Well, lots of fun things as I got to experience in the BI/DW panel at ODTUG. There were some interesting questions from both the audience and Mark (Rittman that is…). The folks on the panel were Michael Armstrong-Smith, Mike Durran (Oracle), Dan Vlamis and me.
Here is a short summary of what I remember from the discussions in a Q&A format. As a disclaimer, I’m trying to explain what people said and therefore I will summarize, consolidate and interpret and hope that it is informative. Additions and corrections welcome!
Q1: In these economic times, can you afford to invest in BI/DW or should you be looking at not investing here?
A1: No, no and more no. As Dan said, this is obviously to be expected from the folks on the panel. Both Michael and Dan agreed that BI is growing and really all about investments to understand things you do not know. Look forward and understand the business. My whole point was about investing in the right things. Focus on what makes you stronger as a competitor. Focus on things like data mining to tell you what you need to know but cannot get from regular BI. Also of course, don’t invest in more conventional storage, but invest in things like Exadata, compression, partitioning etc to reduce storage cost and gain performance at the same time.
Q2: Oracle has acquired quite a few companies and technologies over the past few years, doesn’t this continuous integrating of new things put Oracle at a disadvantage?
A2: As Mark said, while we may thing competitors are better they are really not. And Dan really pointed out that while there are a lot of new things coming into Oracle, the communication around direction is very good in general. My contribution is that while we buy a lot of technology, we also innovate and build things like Exadata allowing us to add value to a storage array.
Q3: What can customers learn for their internal systems from the acquisitions that Oracle does and how it deals with integrating companies?
A3: Well, my observation, first hand, is that we really invested a lot on data quality. Here we built a system that cleanses all customer data and then gets fed into a single instance for all Oracle leads. We now have this down to a science and it saves a lot of effort and money. Mike added that we do a lot on publishing the lessons we learn from implementing BI EE for example at Oracle. Technotes on metalink for example. The other observation Mike had was really about us choosing what to use, and ensuring customers understand how we use products. When do we use OBI EE and when do we use OLAP, just to name an example.
Q4: At very large DW sizes (upward of 50TB) this customer stated that he struggles making external engines to database work, e.g. in-database processing is more effective and efficient with those types of volumes. He quoted Essbase as an example. What is the panels experience and what do the experts think about the future?
A4: Dan took the lead on this and really confirmed that there is definitely some truth to this. While you can add faster pipes etc, you still have to move the data. He did point out solutions like XOLAP for Essbase which virtualize the movement of data. Mark even remembered some 11.2 features he heard mentioned – MDX embedded in PL/SQL calls. So there are solutions coming or in place to not move the data. My point was that, we are seeing more and more integration and push-down of technology to lower levels in the DW environment. We look at Exadata, where we push down to storage, we look at pushing down more complex things. We also see embedding things in chips to make the solution faster. So over time we will see more and more consolidation, and fewer engines in the data center.
Mark did point out that in many cases a separate engine/product is chosen for pragmatic reasons. E.g. who buys the product and what does it do? Is it central IT, or the lines of business for accounting purposes?
Overall the panel did seem to think that in-database is less complex as was concluded by the customer asking the question. Even in departmental solutions his argument was that these systems need to talk to other systems, hence the data movement and integration that is still required.
Q5: Let’s shift gears was pointed out by an attendee. All this technology is lots of fun, but aren’t people trying to solve real problems? So what are you seeing customers asking for?
A5:Michael started off with stating that a lot of customers want to move onto newer products first and foremost, as he deals with Discoverer customers a lot. Price was often an issue he thought. Dan added that on the price issue customers should look at OBI SE One, often a forgotten product/bundle but quite a good product to start with. Mike then brought it back to the question and pointed out the BI Applications as something that offers a pre-built solution. My comments were in that general direction. We are seeing more and more pre-built content, more and more pre-configured systems. On top of that there is more vertical focus at Oracle and at our customers. Generic is not good enough, industry specific and pre-configured is what our customers want. Something like a Database Machine for Telco’s with a Telco data model and pre-built KPIs.
Q6: So when will Discoverer die?
A6: Talk about a provocative one… Both Michael and Mike (from the partner and Oracle perspective) agreed that never was the right term to be used. There are various options, from just using Discoverer to augmenting it with BI EE components to switching over. Dan’s point was really around marketing, he was saying, even if the tool is not going away, don’t expect Oracle to market it anywhere… Mark thought it really was like Forms, yes it is no longer a strategic product at Oracle, but customers are happily using it and slowly moving into Java based development.
Q7: As a close-out question Mark was asking us all for some advise. What wisdom could we share?
A7: Michael stated that he is predicting the demise of the disk drive… all going to solid state solutions within the next year or so… then he went on to say that it is really one of the crucial things in BI to talk to the business, bring business users together and focus on that. Mike’s advised the attendees to focus on ease of use for the end users. Use all the resources out there in blogs and tech notes to allow any user to get at his or her data. Dan focused on doing projects well. Start small, build prototypes and gradually deliver solutions in incremental stages.
My advise? Buy Exadata!!! Mark wanted more, so I gave him “buy more Exadata”… jokes aside, the point is that people should – as I was trying to say on the first question – look for software that helps them push the envelope. Find things that bring you a competitive edge, that gives you more than what average software gives you. That can be Exadata, that can be Data Mining, that can be and probably will be a combination of the above.
So all in all, quite an interesting cross section of questions and opinions, thoughts and ideas. Always fun to be in a panel!