Making Data Simple: Fast data with superhero Adam Storm

Making Data Simple: Fast data with superhero Adam Storm


You need to input and output data at rapid and ever-increasing speeds. How do you keep up? Where should you store your data and how can you optimize it? In this week's episode of Making Data Simple, Al Martin and Adam Storm, IBM senior technical staff member and master inventor, next-generation HTAP architect, sit down to talk about fast data. Adam also covers the pros and cons of different information architectures and the software you can use to help you optimize your data.


00.30 Connect with Al Martin on Twitter and LinkedIn.

00.40 Connect with Adam Storm on Twitter and LinkedIn.

01.17 Listen to Making Data Simple: The Big data problem with Daniel Hernandez, vice president of IBM Analytics Offering Management.

01.25 Listen to Making Data Simple: IoT, patents, and invention with Lisa Seacat DeLucaIBM distinguished engineer, IoT App Factory.

01.30 Listen to Making Data Simple: Making data fun and easy with Caleb CurryYouTuber and IBM social strategist.

03.40 Read "Cloud Analytics: Derive Insight without On-Premises Infrastructure" by Ahmed Fattah.

03.55  Explore and learn about IBMs IoT initiatives.

03.57 Learn more about blockchain

05.40 Learn more about fast data .

10.40 Read "The Lambda Architecture simplified" by Adam Storm.

10.45 Learn more about KAPPA Saphir.

13.45 Learn more about IBM Data Science Experience.

15.95 Learn more about real-time analytics.

16.10 Learn more about Synopsys.

17.50 Learn more about Forrester.  

20.21 Want to know more about Apache Zookeeper? Cick here. Want to learn about Spark? Click here.

21.45 Learn more about Apache Parquet and its uses here

24.50 Learn more about fast data, the Event Store and connect with Adam Storm by eventstore@ca.ibm.com.

Ready to dig deeper? Check out our previous podcast episodes of Making Data Simple.


Al Martin:             Hello.  You found your way to the analytics insight in the Making Data Simple series. Right now, this is a staple with no end in sight. Today, we are going to talk about hyperdata management in a product called (unintelligible) as well.   

                              I’m your host Al Martin, and today with me I have Mr. Adam Storm, who is the next generation hyperdata architect and I if I can say so, master inventor.  Hi Adam, how are you? 

Adam Storm:         I'm good, how are you? 

Al Martin:             I'm doing just fine. Adam Storm, you know you have a superhero name, right? 

Adam Storm:         I hear that from a lot of people, yeah.

Al Martin:             So I've got Albert Martin, which is a nerd name. And you have Adam Storm, which is like a superhero for nerds. 

Adam Storm:         Yeah, it's a good thing that this is audio only because then you can't see how nerdy I am in real life. 

Al Martin:             No, you're a superhero. That's for sure. So are you listening to any of these podcasts? 

Adam Storm:         I have listened to some, yes. 

Al Martin:             What's your favorite? 

Adam Storm:         I like some of the early stuff. I like the one with Daniel Hernandez. That was a good one.   

Al Martin:             So you have listened. So we're getting worse is what you're saying.

Adam Storm:         No, I think I just like the oldies, the oldies and the goodies. I also like the Caleb Curry and I have the Lisa Seacat DeLuca one on my phone read to listen to.  I haven't gotten to it yet. 

Al Martin:             So, you know, we're nearing about 100,000 downloads, and it's only been since late last year. For everybody that's listening out here, I appreciate it. Please rate us on iTunes. Provide feedback.   

                              If your company is doing something interesting, you want to come on, have a chat, you're certainly welcome. We're listening. So give us some feedback. 

                              Okay, master inventor, what do you want to talk about today? 

Adam Storm:         That's a good question. I think we should talk about what's going on in the fast data space, because I think that a lot of people may not be aware of what's going there. I, personally, think it's a very interesting space right now. 

Al Martin:             Here's my first question, though. What's the one step down from master inventor? Is there like protégé inventor, training inventor? I'd like  amateur inventor. Because I never see anybody's business card that says like amateur inventor. It's always master inventor. 

Adam Storm:         Yeah, we like to keep things binary here. Either you're not a master inventor or you are. So it's really  there's no stages. 

Al Martin:             Fair enough. This guy's good at his answers. All right, so let me take a step back. And master inventor, why don't you describe your area of expertise and well that's deserving of a superhero name that you have? 

Adam Storm:         OK. So I've been working at IBM for a while now, probably too long to mention how long. Even though there are a lot of people who worked there a lot longer than I have. But in that time, I mostly focused on data. How to store data, how to make it successful very quickly, how to store it in different fashions.  

                              I worked on PDQ for the better part of a decade and a half, and since then, I've been focusing on what we call the next-gen space. So for those of your listeners who know about (unintelligible). They know about all these new  (unintelligible) IBM to go after that space in an aggressive way.  And that's what I've been working on for the last couple of years.  

Al Martin:             Very good, so we'll get to fast data another  what you said you wanted to talk about. But to that point, as you know I lead hyperdata management and my mission is very straightforward: making data simple and accessible. But in reality, simple is not easy.   

                              And for (unintelligible) data chaos today. I mean there's data everywhere. It's every format structure, unstructured, semi-structured, cloud, on-prem, hybrid cloud. Which brings the concepts like data gravity putting for data and other data so you want to put (unintelligible) the data. But to your point, when you said you want to talk about faster data, data is as large as it's ever been. It's faster than it's ever been.   

                              Now we got ISP to contend with. We have blockchain around the corner. Even that weird (unintelligible) in the business of wrangling data, let's talk a little bit about the problem statements that you're working to resolve in the architectural role. Could you start there? 

Adam Storm:         Yeah, sure. So it's a big space. And like you said I think that data, you know, if you look 20, 30 years ago, data was a fairly simple problem to solve. There was only a couple of ways that you could store it.  

                              A couple of tools that you could use to store it, but your options were kind of limited as to how to store the data and how you can make it accessible. So we had a  it was simple in some respects.  

                              In that if you had a data problem, you know, you need (unintelligible). But your options weren't as wide open as they are today. 

                              So now fast forward 20, 30 years, especially in the last 15 years with the rise of things like (unintelligible) and all of the open source technology that's out there. There's a whole number of ways that you can store your data. The problem is a whole bunch of those ways are not necessarily good ways to store data.  

                              And you see there are a lot of architectures being proposed in open source and in forward-thinking companies that may work for them at the scale that they're working at.  And because they want to leverage open source tooling and they have the staffing to do that, but they're setting up tremendously complex architectures which companies that don't have their resources aren't able to do. 

                              So I think that you and I are both trying do the same thing, which is to take this massively increasing amount of data and make it simpler to manage. So it's a take the complexity that's come into the industry in the last 15 years, and really simplify it at an enterprise scale. 

Al Martin:             Well let me ask you this. Why, when I ask you what you want to discuss, you go immediately into fast data? 

Adam Storm:         I think that fast data presents its own interesting challenges. Honestly, the data is coming in extremely quickly. So much so that we  there's streaming technology as of today that's the predominant way of handling fast data. And what they do is they analyze it small sections and then just throw it away because they can't (unintelligible) it fast enough.    

                              So there's the challenges of bringing the data in quickly, but also as you're accumulating all this data very quickly, you end up accumulating a huge amount of it very quickly. And so you need to handle the storage. Store it in an effective way, but also in a way that you can make use of it.

                              Having data just isn't valuable. Deriving insights from that data is the value. You need to derive insights faster and faster today to get value out of the data.  

Al Martin:             So when you and I have talked about this before, you had mentioned various solution vehicles for this problem like land architecture, cap architecture, modified land architecture. Seems like there's no perfect solution.  

                              And I know we can get in the bowels quickly with some of those, maybe those three areas of discussion. But at least at a high level, can you give me your thoughts?  I mean, is it as confusing as it sounds with all these different architectures to solve the same problem?

Adam Storm:         I think it is as confusing as it sounds. Okay, let's take a step back. So if you look 15  just 15 years ago, there was this some might call it a renaissance open source in the hand of data professionals.  

                              So there was (unintelligible) came out of Yahoo, Facebook open source (unintelligible). There was a whole bunch of LinkedIn open source (unintelligible). There's a whole bunch of open source tooling that's available now.   

                              And a lot of people are seeing this and saying we can cobble this together with a solution with all of these components. Which I think works for a lot of companies especially if you're willing to put in the energy.  

                              But it's tremendously complicated for a shot that doesn’t want to devote a huge percentage of its resources to bare-bones IT and data management. It's difficult for them to take the same approach. 

                              And I think one of the big problems is that if you go on YouTube or you go on (unintelligible) and look at all of these presentations, these companies like Google and Facebook and Netflix and Uber have huge IT departments that are putting together infrastructure to store all this data.  

                              If your company doesn't have the resources or the appetite to do that, there has to be a better solution. And there needs to be a simpler way to store these kinds of large data sets. 

Al Martin:             Well, to your point, I mean, the thing that's really interesting to me is I'm talking to many clients about cost savings and then in some of these architectures that you mentioned, I turn away and then look back. And they've got small armies that they're spending to stand up these architectures. And I'm questioning, now what was your original goal again?   

Adam Storm:         I've seen that a lot, too. Like I talk to customers and especially over the last five years, it seems like there are a lot of larger customers who  when I say larger, I mean large by market cap. And they think well we can just roll our own data solutions. Leverage open source. Build it (unintelligible).  Bring in all the best of breeds of like the Cassandras and so on.  

                              And we'll roll our own architecture. And two or three years after they've gone down this path and have not put anything into production yet, we end up talking to them. And they say this was the wrong approach. We don't have the staffing, resources or the appetite to do all this. And we want an enterprise software company to deliver a solution for us.

Al Martin:             To that end then, what is the right approach?  I know we're going to talk about a little bit of the Db2 (unintelligible), which may be part of the solution here. But back to the original question. What do you think is the right approach? And who do you think wins in the future here relative to the problems that you've outlined? What is the winning formula? 

Adam Storm:         Yeah, so I can start by saying for sure there is no single right approach. There can probably have different data (unintelligible) and there will be slightly solutions for each of them. What I can say is I think the companies that will win in this space are the companies that make simple solutions that solve the broadest stuff that  of use cases.  

                              Specifically, you can't have a separate solution for your fast data and your slow data. You can't have a requirement that people stand up a fast data (unintelligible) and then move the data over time to a slow data or a faster repository. 

                              You mentioned Lambda off the top or a few minutes ago. This is really the main proponent of the land architecture (unintelligible). And use one for pushing in the data quickly and one for (unintelligible). I don't think that works long term, because people don’t' want to manage two stores. So getting back to your original question, who wins? 

                              I think we're now in a race for simplifying the story. And the companies that will win this battle are those that bring simplicity the soonest. 

Al Martin:             A lot of clients, correct me if I'm right or wrong here, but are investing in like the land architecture which, you know, seems like back to the future. I don't know if that's the right words for it. 

                              Yeah, the  you're investing in two different layers if you will, speed layer and (unintelligible) layer is  are people moving away from that? Or do you still see clients heading in that direction and you're shaking your head, or what? 

Adam Storm:         I think that we're starting to see people moving away from that. So (unintelligible) was huge between five and ten years ago. People are starting to see the complexity of it.  We're seeing alternative architecture like (Kappa) and modified (Lambda) that are trying to manage some of this complexity.   

                              But we're also seeing platform emerging from companies that aim to handle the two different stores and consolidate them into a single offer. 

Al Martin:             For our listeners real quick though, we've been talking about (Lambda) and (Kappa). I mean, I know these are detailed architectures, but can you just give a quick overview of what we're talking about here? 

Adam Storm:         Yeah, I can try. So the Lambda Architecture, the principle  the fundamentals of Lambda Architecture are that you cannot persist data and make it queryable at the same time. You can't process it fast enough and make it queryable immediately.

                              So the way they solved this problem is to set up a speed layer which does the data persistence and then there's also a batch layer. That, and data flows into both of those layers at the same time. 

                              Now on the other side, there's an application that's pulling data out of both of those layers. One of the big problems with the lambda architecture is that the application has to know what to pull out of each of those layers. So there's really a few problems.   

                              The first is there's increased application complexity, but then also, there's the data is duplicated in both of those layers. So you have to make two copies of the data and then you have two stores. Two separate sets of infrastructures that you have to maintain. So those two problems. 

                              Kappa, just to talk a little bit about Kappa. They tried to simplify the problem by getting rid of the batch layer. And just using the speed layer for the queries as well. I think the immediate problem with Kappa I mean the obvious problem with Kappa is that it works in some cases where you can have a speed layer that you can query efficiently.   

                              But there are many types of queries that aren't efficient when running on top of the speed layer. 

Al Martin:             Nicely done. So you've been inventing this technology that you (unintelligible) these two (unintelligible). I think it's available now? Customers can rush to get it if they choose to do so if they're listening to this podcast. But does this solve this problem?

Adam Storm:         I think it does. We think it does. And one of the ways it solves the problem is by consolidating the batch and the speed layer into a single offering. So some of the characteristics that we think are desirable in the fast data space that the events (unintelligible).   

                              You can bring data in very quickly, millions of data points per second. The data is immediately made available for analytics and fast analytics. All of the data is indexed, but also we provide a metadata on top of it which allows to query very effectively.   

                              So basically it's a single solution. It's a single solution you can lay down using docker and super (unintelligible). It scales. It's integrated with some of our other offerings at IBM like the Data Science Experience. So that you can leverage first-class data time environment. Lay all of this down at once with a single offering. So there's one thing to manage. 

Al Martin:             Is no one else doing this? 

Adam Storm:         There are people who are doing it. I mean they're  I would say that there are not that many people who are doing it for customers that want to keep their data in-house. There are customers that are  have cloud offerings. Sorry, our competitors who have cloud offerings that solve similar problems. All of them are in beta right now.  

                              And they're being (unintelligible) to those to the marking. But we can see that this is  I think it's on the (unintelligible) we're pretty validated on this approach in the sense that we see our competitors rushing to do the same thing. And so this, to me, screams that we've tackled the right problem. 

Al Martin:             I know this DD2 event store, so that implies DD2 proper. Why is this a separate offering than from DD2?  Just to keep you honest here. 

Adam Storm:         Well one of the reasons why is when we started to undertake this project a couple years ago, there was an appetite among (unintelligible) as a startup model. And so we created a pretty small team. We now have like 15 people on the team, but originally we had around 10. And we worked in isolation from the rest of the organization for 18 months to put this together.  

                              And I think that we saw that there were a lot of benefits to operating in that model. The team became very startup oriented, in that everyone does a little bit of everything. We were at a sprint pace the whole time. We had to validate our findings every six months to executives. So I think it really was a little bit of a, kind of, anomalous experience for IBM. But the success that we brought to the project, I think couldn't have happened if we weren't running in this mode. 

Al Martin:             So let me breakdown some of the differentiators, because I think this will be interesting to the listeners and it'll be interesting to me as well. So you got an event store. It's examining or driving insights to data at an expedited speed, so to speak.   

                              So let's start right there.  In terms of the differentiators, to bring data invest, fast data, even a million entered per second, what if I want 2 million? What if I want 3 million entered per second?  

Adam Storm:         Yeah, for the cluster, when I say a million enters per second, you can drive that on each node of the cluster, each physical machine. But it's a cluster solution mostly for high availability and disaster recovery purposes.  

                              So if you want 2 million or 3 million, you just add more nodes. And the system scales that accordingly in terms of (unintelligible). 

Al Martin:             Lineally? 

Adam Storm:         Lineally, yes.   

Al Martin:             Lineally, did I say that right? That's my Midwest slang, I guess. So all right, real-time analytics. What does that really mean? 

Adam Storm:         It means that you can derive insights from your data immediately. And what I mean by that is, I covered just a little bit, but let me go into it a little bit deeper. So all of the data we bring in is indexed.

                              So what that means is if you just want to pull up the needle in the haystack type query, you want to get one data point. I want to pull Al Martin's record out of the database, that's extremely fast. We also have a synopsis built into the database for those who aren't familiar with a synopsis.

                              The main idea is we sort the metadata that (unintelligible) which allows to determine very quickly which sections of the database we need to access when we're pulling out certain things. So that has the ability to accelerate queries up to a thousand times faster than they would without it. 

                              And all of these factors combined allow us to derive insight very quickly. And must faster than technology (unintelligible). 

Al Martin:             Can you give me an example of quote unquote a (unintelligible)? 

Adam Storm:         I can. I mean principally the (unintelligible) of a space can be thought of as synonymous with the IoT space (unintelligible). The idea is that you're driven more by the events that are coming into the system and acting on those events, persist in those events, deriving machine learning models based on them.   

                              And performing analytics to drive into, you know, anomalous behavior that's going on among your sensors or figuring out an aggregate how things are performing. So really there's a fundamental unit of measurement is the event in the system. 

Al Martin:             How does this fit? I mean I'm very interested in IoT for a number of different reasons, but I think that's where the industry is heading. Or I think there's certainly a lot of opportunity in our very short future around IoT.  Tell me how the architecture fits with IoT. I get the event-driven piece, but give me more information as to who it will be set up. I don't know if you have a use case or something that you can talk to. 

Adam Storm:         Yeah, so we're  I don't want to get into too much detail because we don't have public references yet. But we're working with a lot of customers in the IoT space. And I'm sure you can imagine what that would entail. Customers in manufacturing want to track devices that they have, that kind of stuff.  

                              And the principal problem that they have is the data is flowing in very quickly. There's a lot of data so sift. They want to drive value out of that data immediately. 

                              We were in a  we had a webcast a couple of months ago with Forrester. And Forrester, the big push right now is on fast data and how the value that you derive from your data (unintelligible) exponentially with time. So if you're going to only be able to land that data and derive value from it in an hour a day or a weekend, the data's pretty much useless to you.  

                              So we need to be able to derive  especially in the IoT space, where you need to make corrections in your manufacturing process immediately or you  if you're tracking data that's coming off a (unintelligible), you want to be able to alert customers as the recalls immediately.  

                              So that they can take the appropriate actions. It's important to be able to derive the (unintelligible) very quickly.  

Al Martin:             I don't know. So let me (unintelligible) a few technology questions that are underlying to this technology. I just want to get your input, separate from (unintelligible) and storage capability. Can you do it or no? 

Adam Storm:         Yeah, it can. For those of you who are unfamiliar with the ability to separate competing storage allows you to scale your cluster either based on requiring more storage or requiring more querying processing power. We can do that. 

Al Martin:             What about data availability without duplication? I hear you talk about this all the time. 

Adam Storm:         Yeah, so that's one of the principles of Lambda that I personally frustrates me the most is two copies of the data. And so in the event store we maintain a single copy of the data.  

                              And one thing that I haven't mentioned is that data is stored in the (Apache Parte) format. It's open so if you have other (Apache Parte ) tooling that can query that data, it's available for the (unintelligible) as well.  

Al Martin:             I really don't want this to sound like an infomercial, but at the same time, you know, these questions lend themselves to, you know, exactly the architecture and why would you choose this architecture or otherwise. And back to your original problem statement of fast data and where we started. What's your view on open source?  

                              And if I'm a client listening to this, what should I be thinking at to an open source databases? Some other competitive players that may be out there. I try to make this podcast as agnostic as possible. What's (unintelligible)? 

Adam Storm:         I'm a huge fan of open source.  I mean if you look at the pace of the industries' moves (unintelligible) in the last ten years. It's just tremendously accelerated due to the fact that these large companies are open sourcing huge components of their technology stack.  

                              And I (unintelligible) we're open sourcing left and right as well. And I mean we're so much of spam (unintelligible) at IBM that we've leveraged open source pretty heavily in the (DBQ) events or use of the (Apache Parte) through (Apache Spark), Zookeeper. We use (unintelligible) as well. 

                              So I think that from an industry perspective, if you turn you back on open source, you're in trouble. The pace that open source is progressing is much faster than you'll be able to progress technology on your own. So it's more at your own peril. 

Al Martin:             So competitive players in this space, can you talk to them? You referenced them a little bit. I don't know if there's

Adam Storm:         I can. I think there's a few. The one I think that's most similar to what we're doing is Databricks Delta. They are, as far as I know, still in beta. But they have pretty nearly identical value proposition to what we have. There's other companies that are more build your own stacks.  

                              They're using (unintelligible) something like we have using (unintelligible). And probably spark and there are companies that are doing that. By pushing data on Cassandra and then using the data stacks and sparks connector to pull data out. 

                              And then there's some other players that I think have similar value propositions but are doing it at a difference scale. So companies like (unintelligible) are aiming to do transactions events (unintelligible) and also Snappy Data. 

Al Martin:             One thing before I move on. You mentioned (Apache Sparks) and (Apache Parte). What does those technologies have to offer the solution? 

Adam Storm:         So (Apache Parte) is the data format. All of the data we prescript is persisted in (Apache Parte).  (Apache Parte) is (unintelligible) format (unintelligible) data engineers know is the most efficient way for accessing data for analytics. (Apache Spark) is a run time engine that's (unintelligible) but actually (Apache Parte) did a very quick. 

Al Martin:             I got it.  So where is the future headed? What's the future of this technology? If I'm listening, what should I be thinking as I move forward? And I guess I would have a dual part question. I'd say what is the future?

                               And I'm a client, what also is the trigger point that I should look at this technology or otherwise? I don't know, I just  

Adam Storm:         I think that from a client perspective if you have a problem with  a fast data problem, if you have data that's arriving very quickly, if you  especially if you tried to solve that with a (unintelligible) on your own and you've realized that it's not really tenable long term. I think it's something to consider.   

                              In terms of the future, I think that the future holds a lot of what we've seen over the last 10 years, continuing along that path. I mean we are trying to  this stack has gotten so complex that companies are now identifying that they want to simplify it.   

                              And it's my guess is there will be more offerings like this down the road where stacks are collapsing into single offerings. So that the customer doesn’t have to do this on their own. 

Al Martin:             So that's your future statement then. 

Adam Storm:         Well that's my hope for the future. It may not go there, but my hope for the future that this in place. 

Al Martin:             What is, in terms of a future statement with event store specifically, what are the next few things you're looking at working on in terms of advancements, added features, whatever? Any thoughts? 

Adam Storm:         Yeah, there's a lot of stuff we're working on. I'm not sure how much we can talk about. But I can tell you that, broadly speaking, we're working on improving our (unintelligible) to a bunch of different data sources. You need to be able to push the data in and the more ways we allow customers to push data in, the (unintelligible) of the aperture of what we're working on.  

                              The other thing is we're working on some advanced ways to manipulate the data. And the (unintelligible) is so that customers can maintain it over their life cycle. That's probably sufficiently vague to not get me into trouble, but hopefully give people the impression of what I'm talking about. 

Al Martin:             So look, I think we tackled the topic of fast data. You know, there's a lot more things to talk about in hyper data management. And I know your role, you know, in architects across the whole umbrella there and maybe we'll have you back from that perspective. But as it relates to fast data, anything else that we missed?  I didn't ask if you want to get to our (unintelligible)? 

Adam Storm:         You know, I think we covered it pretty well.   

Al Martin:             If I'm a client, where can go sign up? 

Adam Storm:         That's a great question. So we have a website. It's www.ibm.biz/B-I-Z/eventstore.  All one word. And if you go there, you can download. We have, principally the event store in an enterprise offering. So it runs on that (unintelligible).   

                              But we have a developer addition which you can download and run on your Linux, Windows or Mac laptop and get a feel for really the value proposition. The integration for (unintelligible) experience and the ability to get it quickly and (unintelligible) immediately. 

Al Martin:             Very good, so, you know, we're in the community is a little bit dangerous. So where can the community get a hold of you if they wanted to talk to you more about this? I mean I presume you're on LinkedIn, Twitter, that kind of thing? 

Adam Storm:         Yeah, the best way to get in touch with me is probably to call you Al directly, I'll give you  and then you can just forward their requests on to me.   

Al Martin:             Very nice. 

Adam Storm:         I'm on Twitter and I'm on LinkedIn if you need to get in touch with me. You can contact me there. 

Al Martin:             Adam Storm, master inventor. 

Adam Storm:         We also have an email address for event store directly. It's eventstore@ca.ibm.com.   

Al Martin:             Adam Storm no.  So all right, so I got two more questions for you.  What do superheroes, master inventors do for fun? 

Adam Storm:         The most fun I have is spending time with my family. So I spend a lot of time with my family and my kids. I try to eat dinner with them every day.  And I drive them to school and pick them up.   

                              I spend a great time with them on the weekends. The other things that I do, I'm an avid runner. I run three to four times a week. And yeah, that's (unintelligible). 

Al Martin:             How far do you run?  You run three, four times a week. How long we going here? 

Adam Storm:         Maybe I shouldn't say. I don't know. I'm a little bit (unintelligible) with my running. So I'll run for two to three hours at a time sometimes just to clear my head and get in a good exercise. 

Al Martin:             Are you clearing your head or are you inventing the next big thing with event store or anything under the sun under (unintelligible)? 

Adam Storm:         It's  we can make light of it, but exercising for long periods of time like after you run for, you know, an hour, an hour and a half, two hours. You sort of enter this trance like state where all your head is totally clear and you can (unintelligible) get new ideas.  

                              So I do do a lot of thinking while I run as well. 

Al Martin:             It's like mediating. I got you. I thought you did some Frisbee football or something like that. 

Adam Storm:         I do play ultimate Frisbee as well although I haven't played for a couple of years. But sport-wise, that's the organized sport that I think I'm most. 

Al Martin:             You cannot call it ultimate Frisbee unless there's tackle involved. No tackle? 

Adam Storm:         There's not tackling. I'll send you some videos. You'll see. It's more  

Al Martin:             Oh, all right. You're going to impress me. I'm with you. Look how do you keep up with the latest trends? That's what I want to know. Sometimes I ask folks, you know, what book, they're reading. And you're welcome to do that as well, because I will take that down.  

                              But look, now do you keep up with the latest trends of all these big changes? Whether you're talking Kappa Architecture, Lambda Architecture. Holy smokes. What's your secret? 

Adam Storm:         This maybe (unintelligible) an answer, but I would say the internet. I mean all the answers are out there. There's a what I like to do is like to take an (unintelligible) and the I get directly to my inbox all the latest (unintelligible) that's going on.   

                              That helps me keep up with, especially the industry is just moving so fast. Every day, my inbox full and it's actually (unintelligible) companies X, Y and Z. 

Al Martin:             Set aside a certain amount of time a day to review this information or what?

Adam Storm:         I don't set it aside, but I just get it in. So I probably spend between one and two hours a day analyzing what's going on in the industry and keeping up with the best solutions that are coming up. 

Al Martin:             Are you listening to it in your four-hour run? 

Adam Storm:         I listen to podcasts too. And I love listening to podcasts. Although I like to decompress when I listen to podcasts, so I don’t listen to that many industry podcasts, just this one. 

Al Martin:             Yeah, just this one. Yeah, that's nice. I think we'll end on that one. Nicely done. Well, everybody, the legendary, the superhero, Adam Storm has been with us today.  Look him up and thank you, man. This is a good discussion.  I appreciate everything you provided here. 

Adam Storm:         Thank you. No problem. 

Al Martin:             All right. I'll talk to you next time, Adam. See you.