Ron Powell, independent analyst and expert with the BeyeNETWORK and the Business Analytics Collaborative, interviews Rob Armstrong, data and analytic enthusiast for Teradata. They discuss the importance of managing the new types of platforms and data by continuing to follow the best practices for data governance and management.
Rob, in past years you have had intriguing themes, and one that weíre going to be talking about this year is leveraging Teradataís past to enable your future, but it could also be called: In with the new, but not out with the old. Whatís the story behind this?
Rob Armstrong: A lot of people are looking at the shiny new technologies and trying to leverage them to get into big data and the new analytics. I remind them that they may have a lot of value in their existing environments that they havenít quite yet tapped. Theyíve spent a lot of time, money and sense of humor building data warehouses that may not be fully leveraged. So letís not throw everything out. But, yes, letís bring in these new tools and capabilities to supplement them.
Can you give us an example of this old and new theme in analytics?
Rob Armstrong: Certainly. Let me give you a quick example. Itís about using new analytics against your old data. I donít know many who have ever done text analytics in SQL, but I know very few of them do it a second time because itís very hard to do. Now we have new analytic models that make it easy to do text analytics. So one of our clients decided to take all of their repair bills and look at all the write-ups. They looked with text analytics, and they found fraud from people adding repair parts that didnít match the repair. The example was a shipping crate. A shipping crate is a defined size, and they found repair orders that had two different sized doors going on them Ė one a 9 foot and another a 10 foot. Obviously, they repaired one door, and then somebody later put a second door on the repair order and said, ďrepaired left door.Ē They were able to save a lot of money by discovering fraudulent charges to repair orders by doing new analytics. †
You see the same thing with pathing and affinity and propensity Ė all very hard to do in SQL, but all needing to be done against the data you currently have.
How about an example from the data side?
Rob Armstrong: On the data side, we have new data that needs old analytics. So we have a lot of data that comes in Ė sensor data and IoT data Ė that is fairly relational in nature. I can use my SQL analytics and complement what I am currently doing to make better and more targeted actions. The example I have here is spoilage in fresh produce. In the past, weíd just do some SQL analytics to see what got thrown away and how long it was there. But now we have sensors on the trucks to determine the temperature, the pressure, and the geographic location. And what they actually found was that the average temperature of the truck is okay, but there are hot and cold spots in the truck. So Iím using sensor data Ė still with my old SQL-type analytics Ėbecause it fits the need and directs the action more appropriately.
Recently Teradata has talked a lot about Teradata Everywhere and embracing the cloud. We go from the on-premises Teradata to now either managed, private or public cloud, which is relatively new. Where is the ďoldĒ in that situation?
Rob Armstrong: Itís actually kind of fun because the cloud isnít new. The cloud has been around for a long time. One of my customers said she had to know about the cloud. And I asked her, ďWhereís your physical Teradata box now?Ē She said, ďI donít know. Itís in one of our buildings someplace.Ē Itís in a cloud then, right?†
The idea of a cloud is very valuable, and it brings a lot of capability, but it is just a platform. At the end of the day there is hardware someplace, but itís about the instantiation and the cost of using it that is new. The old is what we bring to the table: a database that has an optimizer, that has parallelism, that has workload management, and has referential integrity. Itís that software in the cloud thatís important. We ran a test at Teradata Partners that shows Teradata in the cloud versus other databases in the cloud, and the results showed how much better Teradata can perform. It is Teradata, not the cloud, that is making the difference.†
Iíll finish by saying that the big part of what Teradata brought to the table is that we have based our technology on parallelism. We didnít base our parallelism on technology, so it works wherever we want to put it. And that is how you get Teradata Everywhere.
So you are using the same version of Teradata, regardless of platform, right?†
Rob Armstrong: Yes, and the best part about that is that if I want to do test and development or I want to do beta testing or I want to do some other type of special processing, Iím using the exact same tools I am going to use when it goes into production Ė because I wonít know that itís someplace else when it goes into production. ††
Awesome. Another big area other than the cloud is the Internet of Things (IoT) or Ė as Teradata frames it Ė the Analytics of Things. What are the items we can learn from the past in this area?
Rob Armstrong: There is the Internet of Things and Analytics of Things, but Iím actually going to coin a new term here: Action on Things. Data is nice, analytics is better, but action is what you need to achieve. We need to be able to take an action on all of this.†
Many people come to me and they say, ďRob, what are the best practices in managing big data and IoT data?Ē I ask them about their practices of governing their traditional little data. When they say they donít do it well, I give them the bad news. While some data may be new, data management is still the same. It is still necessary to be sure the data is governed.†
One example is the sensor in my car. I have to know what car it came from Ė thatís a VIN number. I have to know what sensor it came from Ė thatís the sensor part number. And then there is a lot of information that I really donít need to manage. But if I donít manage the VIN number and the sensor number, there is no way to connect it with anything else to drive my analytics and take action with the owner of the car. So I still need to manage. I just donít manage to the same degree. But the best practices still apply. You manage data to the value it drives. †
Rob, I know weíve covered a lot. Are there any last comments youíd like to share?†
Rob Armstrong: Iíve been in this industry longer than I care to admit, but itís several decades now. I keep hearing that Teradata is going through a new paradigm and that theyíre a new company. Yes, we have embraced a lot of the new, but what has Teradata always been? Teradataís mission has always been data management and analytic enablement. And thatís exactly what weíre doing Ė taking our intelligence, experience and best practices from the past and moving it forward.†
Iím an analogy guy, so Iíll close with one more analogy. When I went to get my house rebuilt, we had a contractor who had over 30 years of experience. When he started out, he used a hammer and a handsaw. Now he uses a nail gun and a table saw. But he has 30 years of experience renovating and rebuilding houses. Iíd much rather go to him than the guy with the shiny new laser level that has no clue about renovating a house.†
The good news is that my 30-year contractor now has a shiny new laser level and he knows what to do with it. And thatís Teradata. We integrate the new toys with the best practices that weíve always been delivering to our customers.†
Very good, Rob. Thank you so much for sharing this enthusiasm for Teradata with us.†
SOURCE: Manage Data to Drive Action for Analytics
Recent articles by Ron Powell