Wednesday 22 June 2016

Podcast Interview with Aaron Wallace, Pitney Bowes

One of the coolest use cases for a graph databases that has seen a big uptake in the past couple of months and years, is Master Data Management. Kind of a vague term, but Wikipedia defines it as
master data management (MDM) comprises the processes, governance, policies, standards and tools that consistently define and manage the critical data of an organization to provide a single point of reference
Sounds about right to me. Lots of customers have been developing their own solutions to their specific MDM problems, but some people have been thinking about this in a generic, generalized way. Like for example our customer Pitney Bowes. They have been early adopters of Neo4j, and have been articulating that vision for the longest time: see this video from 2014 (including other folks from UBS, TomTom, eBay/Shutl as well), and more recently a recording of a talk that Aaron Wallace, one of Pitney Bowes' product managers and my guest on today's podcast, did in 2015 at GraphConnect.

Here's Aaron's talk:

That should give you a good feel of what we are on about here - so time to have our Podcast conversation - here it is:

Here's the transcript of our conversation:
RVB: 00:02 Hello, everyone. My name is Rik Van Bruggen from Neo Technology. After quite a long pause after GraphConnect Europe, here I am again recording another podcast episode. Today, I'm joined by someone that I've been looking forward to speaking to for quite some time, Aaron Wallace from Pitney Bowes. Hi, Aaron. 
AW: 00:22 Hello, Rik. How are you doing today? 
RVB: 00:24 I'm doing really well. Thank you for joining us. It's been a bit of an [chuckles] exercise in planning to get both of us on the phone, but here we are. Would you mind introducing yourself to our podcast listeners? 
AW: 00:39 Sure. Hello, everyone. My name is Aaron Wallace. I am a product manager with Pitney Bowes, specifically within our customer information management line of business, where we offer a product to solve broader problems for information management, master data management, data quality amongst a number of other use cases for our customers. 
RVB: 01:02 Pitney Bowes to me was one of those hidden gems of American industry, I suppose. You guys do lots and lots and lots of cool stuff, but most people might know Pitney Bowes from a specific type of product - the postage machines
AW: 01:21 Yeah, absolutely. That's certainly where our history is. It's close to a 100-year-old company. Certainly has its roots in mailing and postal solutions, but pretty large software division that's really grown up over the last 15 or 20 years, I'd say. 
RVB: 01:36 Absolutely. You know what? I've seen some of your work and also what you guys do with Neo4j. It's pretty, pretty cool stuff. So would you mind telling us a little bit about how you got into that, Aaron? How did you guys get into the wonderful worlds of graphs? 
AW: 01:53 Sure, it really all relates back to our decision to enter the master data management market, which goes back probably about six or seven years ago - something in that range. We certainly became aware of a trend as a company that had been offering solutions around data quality, primarily in the initial years of selling our platform spectrum. We decided at that point that we needed to get into the master data management market because really, what you see is a trend where folks really want single solution, single platforms or at least solutions from one vendor to solve the broader range of information management use cases. 
AW: 02:36 So when we made that decision, we looked at the market. We looked at a lot of the existing competitors out there. We listened to a lot of our customers, prospective customers, and certainly heard a very consistent theme as it related to-- some of the issues related to the waterfall methodologies that get employed in a lot of MDM projects. So we decided being a new entrant in the market, that we really needed an opportunity to do something a little bit different, a little differentiated. So that's when we started really looking at graph and the application it could provide for solving master data management use cases and really, how it could address some of these pain points related to agility that we feel are really tied a lot to the way many existing solutions lean very heavily on relational technologies and almost canned models for solving problems for their customers. 
AW: 03:33 Just hear a lot about very extended timelines, very large budgets, and in the end, solutions that don't necessarily meet the requirements of the business user. We related that back, in many ways, to a fundamental issue with agility in managing repositories for master data. So that was what lead us to graph, and then ultimately, what lead us to Neo is a key piece of our solution. 
RVB: 03:59 That's really, really cool. Neo4j has been part of Spectrum probably for a couple of years now. You guys have been deploying it at quite a few interesting customers, right? 
AW: 04:10 Yeah, absolutely. We've have had several customers as the product has grown up, and we've really gained a lot of traction out there in the market place. A number of customers that are doing very interesting things, solving problems across different verticals - financial services, retail, OEM-type models. Really again, solving those kind of issues, dealing with multiple data sources. Many times when you look at information management or master data management in the enterprise, you're talking about hundreds of different data sources that need to be combined to drive 360-degree views or multidimensional views of your customers. 
AW: 04:49 Another aspect of it that really fits well in terms of graph is the business-user focused modeling paradigm that we're able to really leverage as we get spun up in these projects with our customers, really being able to bind together the IT side of the house and the business side of the house as the initial models and requirements we developed. We find that the whiteboard style modeling that graph is really great for really lends itself well to getting these things off the ground quickly and delivering value back to your customers very quickly. 
RVB: 05:21 Normally, I ask people, "Why did you get into graphs?" But I think you've already answered it [chuckles]. It's all about flexibility, the agility, the modeling. Those are the three main themes, I'd say--? 
AW: 05:35 Yeah, certainly. As you know, highly connected data is a very much a sweet spot up for graph when you want to understand relationships. I think the importance of that is becoming just more and more evident as it comes to managing customer information and really managing the customer experience as a business, as you have data sources like social, mobile, like some of the sensor data that we're looking at now with Internet of Things really coming to the front as another element of the multidimensional view. That's another area where it's really great at modeling that and also performances. If you want to do a friends of friends of friends type query in a relational system, that can get really complicated really quickly. It can also break down its scale very quickly. So that was another key aspect for us as well. 
RVB: 06:26 Absolutely. Is this product that you guys have developed-- Spectrum, is that something that is specific to a particular vertical or is this something that you use in all kinds of different--? 
AW: 06:40 No, not specific to any one vertical. We certainly have verticals that we focus on from a go-to-market perspective, but it's applicable across a number of verticals. I would say the biggest one for us these days relates to financial services and insurance. Financial crimes and compliance is a big area for us as well. Another area for the graph, as you know also, is fraud detection. So we're working with a couple of key customers right now to build models for advanced detection of these kind of events and activity within a network to really get out front of some of these regulatory requirements, which as we know - from everything we see in the news - can be a really big deal [?] a lot of these organizations. 
RVB: 07:24 I can tell you, ever since the Panama Papers broke, we've been talking to a lot of people about using the graph for fraud detection. It's been a very [active?] conversation. Very good. So where do you think this is going, Aaron? What does the future hold, both for the graph industry - I would say - for Pitney Bowes and Spectrum? Where is this going, you think? 
AW: 07:50 I see a number of really interesting applications for us - some things we've done in recent releases and things we're going to be working on going forward, again in the general area of our marketplace. One thing we've done recently is kind of a mashup of what we've done with Neo, in terms of the graph along with a federated virtualized engine we can provide to support more of what's a typical registry or hybridized pattern in MDM. Where you don't necessarily need to store all your data centrally within the repository, but rather you can store pointers on certain nodes and effectively design virtualized nodes, entities, and also virtualized relationships between those things. So we're really excited about that. I think that's just another aspect of enabling that kind of agility. 
AW: 08:41 You can start with a smaller set of data that you're mastering centrally and maybe reach out on that registry pattern for other things. Metadata, at the enterprise level, is probably the other key thing I would highlight from our perspective. In many ways, even as we see competitors starting to catch on a little bit to the message we've had out there for a while, we're doubling down on a lot of the stuff we're doing around graph in particular. Looking at areas like enterprise metadata management, and managing and querying unstructured data as well. You alluded to the Panama Papers there. We've tracked that very closely and have actually built some pretty interesting demonstrations using our product that deal with the unstructured aspects as well as a graph, visualization, and query aspects. So we're excited about some of that as well. 
RVB: 09:33 Super cool. Well, I think we'll put some more links to Spectrum and to Pitney Bowes on the transcription website of the podcast so people can find more information as easily as possible. I want to thank you for coming online, Aaron. It's been great talking to you. We want to keep these podcasts fairly short, so we've covered a lot of ground very quickly. But thanks a lot, and I hope we get a chance to meet face-to-face at some point.

AW: 10:02 All right, Rik. Thanks a lot. I appreciate the time. 
RVB: 10:04 Thank you. Bye-bye. 
AW: 10:05 Bye.
Subscribing to the podcast is easy: just add the rss feed or add us in iTunes! Hope you'll enjoy it!

All the best


No comments:

Post a Comment