New York Stock Exchange at IBM IOD 2011: The benefits of using IBM Netezza
Emile Werr, of NYSE Euronext, joins the IBM Break Free forum at IOD 2011 to discuss the benefits of using IBM Netezza. (See more from Emile in Part 1: The challenges of managing your ever-increasing data footprint).
- Transcript of Video
00:56:59 EMILE: When we went to Netezza, we couldn’t even finish our surveillances in a 24 hour window, so we were actually always trying to play catch up. And we had a lot of changes happening at that time. We were, you know — NYSE’s concept is a fast and slow market, which means that there’s constant changes in the rules, trading, and types of activities that you need to monitor.
00:57:19 So not only was our complexity, in terms of the business driving a lot of complexity, and the volume was also driving complexity. When we started utilizing Netezza, we were able to run these analytics in, I wouldn’t say in seconds obviously, but we were doing it in minutes and in worst cases two hours to finish an entire job that never finished in 24 hours. And that was with very little tuning.
00:57:42 The only thing you really need to know when you’re utilizing an appliance like a Netezza is right size; make sure you’ve got the right size for the footprint of data that you need to put on there. And distribution; make sure that when you create your tables and you understand how your data needs to be distributed that there’s no skew. If you have those fundamental basic concepts down, you’re going to get out-of-the-box good performance.
00:58:05 The other thing that I want to highlight is a lot of folks are talking about the complexity in the application tier. We’re big promoters and proponents of proper data architecture and normalized data structures. Because when you do that, you actually have a more flexible and agile environment. A lot of time you can’t do that in traditional systems.
00:58:25 A lot of our success is because we were able to do that on top of Netezza. And that means when business changes, we can change our underlying schemas from a normalized architecture as opposed to constantly rebuilding tables. So those are all parts of, I think, why we were able to sort of leverage and adopt MPP. And we’re continuing to make advancements in that space.