How to Select a Big Data Solution: Video Roundtable

See transcription of video discussion below.

For businesses looking to deploy a Big Data solution, for the first time or as an upgrade, the choices are numerous. Which key factors are most important for maximizing competitive advantage? Is it depth of feature set, ease of use, cloud-based, in-house? What is the state of the Big Data market, and how can businesses leverage knowledge of the current landscape to choose the best solution? I’ll interview three Big Data thought leaders about these and other key questions. Please join us for a lively discussion.

	Dan Kogan, Director, Product Marketing, Tableau Software
	Satyen Sangani, CEO, Alation
	Abner Germanow, Senior Director of Strategic Marketing, New Relic

Transcription – How to Select a Big Data Solution: Video Roundtable

J. Maguire:

Hi, I’m James Maguire, Editor of Datamation, and our topic today is how to select a big data solution. What you as a business should be thinking about as you make that decision. To talk about that we’ve got three great experts, including Dan Kogan, Director of Product Marketing for Tableau Software. Hello to you, Dan.

D Kogan:

Hi, everyone.

J. Maguire:

You’re up in Seattle, I know, right?

D. Kogan:

I am, yeah. It’s a beautiful hot sunny day here. So rare.

J. Maguire:

I was up there several months back. I couldn’t believe it. Seattle is like the ultimate boomtown. Even more than San Francisco. It’s like cranes, everything’s going, it’s enormous energy in Seattle right now.

D. Kogan:

Yeah, we’ve got a couple software companies down here who are building a lot of stuff in the downtown.

J. Maguire:

Yeah, I don’t know if anybody’s heard of them.

D. Kogan:

Yeah, a lot going on here.

J. Maguire:

Also with us is Satyen Sangani, CEO of Alation. Hello to you, Satyen.

S. Sangani:

Hey James, how are you?

J. Maguire:

Good, good. I know you were just recently at the Davos Forum. That’s pretty sexy. Were you like hobnobbing with Warren Buffet? Or, what was going on?

S. Sangani:

Yeah, Warren sought me out, as did every other Fortune 500 CEO that was there, and world leader. Not quite, but that’s what I tell everyone. I was there and I met some amazing people. It was an incredible event and one of those things that I had always wanted to do and we got the chance to do it because our company has been doing great work and I get to take all of the credit and go to the events.

J. Maguire:

Also with us is Abner Germanow, Senior Director of Strategic Marketing for New Relic. Hello to you, Abner.

A Germanow:

Hey, how are you?

J. Maguire:

Good, and you are there in Palo Alto, i know, so you are right in the thick of Silicon Valley.

A Germanow:

Right in the thick of it.

J. Maguire:

Fantastic. So, let’s just say this. There’s a hypothetical company out there and they want to get onboard with big data, or maybe they’re a little bit on board now but they’re kind of confused that their competitors are very much on board. So, say they come to you in total confusion, befuddled, and say, you know, how do we start with this? How do we buy a new solution? How do we select a new solution among the many many confusing choices out there? Dan, what one piece of advice would you give if a company comes to you with that?

D. Kogan:

The first thing I would, not tell them, but probably ask them, is what is it they’re trying to do? What does big data even mean to them? Really understanding their use case. I find, you know, at Tableau we support companies from really large to really small. I talk to a small business and they say they want to do big data, but really what they’re talking about several thousand rows of information that might have been sitting in Excel before and they’re outgrowing what they’ve been able to do with that tool, but it’s still very much structured data, it’s  the types of things IT departments and companies have been working with for a very long time and there’s nothing to do with, you know, we’ll call it log files or streaming data or IoT or anything like that. It’s very traditional, but to smaller businesses it’s much larger volumes of data than they’ve dealt with before. So, the answer is different to a manufacturer of jet airplanes, for an example, or engines with sensors sending terabytes of data over the course of flight and trying to make sense of that. So, the solution really differs, but it’s really understanding what type of data you’re talking about and what the use case is and then I guide them from there.

J. Maguire:

Satyen, how would you start off that process? If someone came to you confused, how do we choose a big data solution?

S. Sangani:

Yeah, I think I dovetail off of what Dan said. You know, I’m a big believer of Simon Sinek start with the ‘why’. I think that’s always helpful. I think that… you know, the confusing and I think the amazing opportunity with analytics is that there are tons of analytic vendors and companies out there. Analytics means a lot of different things to a lot of different people because it’s trying to optimize and really make more efficient the process of human thinking and human reasoning. Many human cognition. And so, there are tons of tools that just do that in totally different ways. So the ‘why’ is very critical  because the why guides the thing that you might look to solve and what your problems are who your users are and where you’re trying to go. All of those questions could lead to some very very different answers.

I think the thing that I would probably start by saying that one-size-fits-all tends not to be the answer. So, that is probably, I think, the best piece of advice. There’s so many different tools out there so being able to just make sure you understand what you’re trying to do and why you’re trying to do it is absolutely critical.

J. Maguire:

Abner, what’s your take on this?

A Germanow:

I would totally agree with Dan and Satyen. That’s what data do you have, what are you trying to solve for, and the one that we haven’t talked about yet is who are you trying to empower?

J. Maguire:

The sales people or the…

A Germanow:

Yeah, and this goes back to sort of like what does it mean to you? What is the problem you’re trying to solve for? Because if you’re trying to solve for, say, a data scientist problem, that’s going to lead you down certain paths. If you’re trying to empower a sales team or a marketing team or both or you’re trying to drive collaboration among a group of people then that’s going to drive that process.

The other thing that I might think about is how unique is that problem. Is that something that is unique to your company? Or is that something that many companies come into solving for?

J. Maguire:

You know that never fails to amaze me? This is sort of a tangent so please excuse me, but gleaning insight from a big data solution continues to be somewhat still hunch related. I mean, you say the numbers don’t lie, the data don’t… let’s go by the data. Let’s forget human emotion, now we’re going to use the numbers. Yet, it still seems that human judgement and human gut instinct continues to be a role in how we deal with big data.

Dan, am I making that up? This is a tangent. Does this make any sense to you? Is it still human emotion or are we really listening to just the numbers these days?

D. Kogan:

Well, I mean we need humans to interpret the numbers. That is our core belief here at Tableau. People are apparently curious, and they have questions about their business or about whatever it is that they’re doing, and there’s data that can help answer those questions. But it is only the people that are the experts of their domain, and of that data and therefore itself and what it means that are able to interpret it. I see machine learning becoming incredibly popular for many use cases, but it’s very hard to just say machines will be able to generate all of the answers to our questions and we will never need to think again. At the end of the day, the tools we’re talking about and the things we’re talking about that we’re trying to get to is to empower people to be curious and to scratch that itch and to find new insights and to keep going and keep digging deeper. So it’s very hard to map. It’s all in the human brain. It’s different to each individual, so I don’t see how we automate that process and answer all of the questions that the human brain has. We’ll call it the unknown unknowns.

J. Maguire:

Right. So what about… getting back to this hypothetical decision process. It seems like part of the confusion in the big data market is that there’s more than one model. I mean, when I go out and buy a car it’s going to have four wheels and an engine regardless of whether it’s a Jaguar or Toyota or whatever, but if I’m going to go out and buy a big data solution it could be any number of models. It could be in-house, it could be buy it as a service, it could be working with consultants, or some permutation of those three. Is there a model that fits… Do you have a bias in favor of one model? One model is going to be far better going forward? Satyen, what’s your take? Do you have a prejudice in that view?

S. Sangani:

Yeah I do. I think that Dan was… I think the right sort of adjective here is curiosity. I think often tool technology selection gets in the way of just getting the answer. Analytics is fundamentally sort of an iterative process. You get one answer and you get ten more questions, and you get another answer and you get ten more questions, and you get another answer and you get ten more questions after that. I think getting started is really the critical idea. If I could say one thing to any individual who’s starting a journey where they’re going from literally zero analytics, or even lots of analytics, to more analytics. It’s really about acceleration, because when you’re getting to a place where you’re sort of creating an inquisitive culture and people are able to answer more questions, we will do the things to reduce the barrier to asking more questions. That’s the kind of learning environment that you really want to build and where the power of analytics lies. And so one clear evaluation criteria would be what’s going to get you going the fastest, and what’s going to accelerate your ability to get to that insight more quickly. That’s both in terms of the ongoing… the implementation of that idea, you know, putting that technology forward whatever it might be. But also, just on an ongoing basis, getting users up and running, getting users to use the system. All of that matters. Cost of entry is really critical because, you know, the last thing you want to do is build it by technology that has every capability in the world, but no ability to be used by human beings.

J. Maguire:

Well, is that another way of saying that you should always start of as big data as a service? Or, are you saying, no, actually, longer term you might as well build it in-house because you’re going to need it in-house? Have you come down in favor of a model in particular?

S. Sangani:

No, I personally don’t. I mean, I think that you could get going with big data as a service, you could get going with big data on premise. I think that is sometimes and orthogonal solution. I mean, if your data lives in the cloud and you’ve got a clear cloud framework and understand how that’s going to work then go with the cloud. If you’ve got all of your data on-premise and you’ve got a ton of Excel files, then go on-premise. So, it really depends on what your strong point happens to be, but optimize for speed. Speed is going to be a different answer for one person with one set of use cases than it is for another person with another set of use cases.

J. Maguire:

Dan, do you favor one model?

D. Kogan:

I agree 100% and to me the is about optimizing for speed like Satyen talked about. When we’re saying big data what we’re talking about are typically augmentations to existing types of data and data stores. So, you had data coming from your systems of record, from your ERP systems and your transactional databases and that structured data that’s sitting in some sort of relational database of rows and columns, essentially. And then you start to add new types of data. Again, that can come from sensors or CCTV or weblogs or all these types of things, and maybe it doesn’t fit so well into the relational structures of your existing infrastructure, but it’s augmenting what someone already has and someone is already using.

So, mistakes we see companies sometimes make is getting behind this big data initiative and going completely dark from their end users as they build everything out, and then dropping it all in at once. But, what I see as a more successful approach is just slowly augmenting and moving over and then adding to the core of what they already have, so there’s more growing in the backend than IT is standing up, but to and end user they’re just now seeing more data and having more input sources to interact with and therefore more questions to get answers to.

J. Maguire:

Well, those companies that go dark in the middle of it, I assume, are the kind of companies that are building it in-house because it’s that elaborate of a structure. If they’re hiring it as a service they’re not going to need to go dark in the middle of it.

D. Kogan:

Yeah, if they brought in… or whether they’re doing it with their own in-house IT or engineering team or they brought in a third-party consulting firm to do it, there’s still time it takes to develop the infrastructure and the different components. And there’s just a lot of choice in different things. If we look at how many components around the Apache Hadoop ecosystem and around Spark and around a number of this big data-related open source projects, it can be very very overwhelming to understand what to use for which use case and what the right building blocks are, essentially. So, whether or not you have that knowledge in-house or you bring that knowledge from somewhere else, there’s still typically time taken to put that together and build the right backend.

J. Maguire:

I guess I keep waiting for someone to say, and maybe Abner, you’re my man to say this, for someone to come out to say, ‘Oh no, don’t ever build your solution in-house unless your Facebook or Google or someone. Building your solution in-house is a fool’s errand and you’ll be outdated in five years anyway, so it’s all about as a service.

A Germanow:

Yeah, so part of me believes in that, but it depends on the use case, right? So, if you have an existing infrastructure. If you have a ton of data that you already gather and parse and analyze in a variety of different ways, then augmenting to that existing system is a useful path.

The other angle that I see is that people have solutions and problems that they’re trying to solve for and they have…

J. Maguire:

Does your wife have an opinion?

A Germanow:

My print out. The thing I was trying to print out earlier finally did.

J. Maguire:

Oh good. You got the data. Alright.

A Germanow:

I got the data. You have this world where when you go try to solve for particular problems, in our case it’s software data, so what happens when you have an application and you’re trying to understand what’s happening inside that application. In a prior life I did a project around sales and marketing data where we had some of the data, but we looked to a service provider for a set of other data and the ability to augment all of that and make sense of it and bring the context to light, and so you do have this situation where, to Satyen’s point, how do you go fast, you also want to go with how do you reduce risk. And today you have existing tools that in some cases you’re not using the capabilities to their fullest extent. In other cases you have situations like let’s say you’re a small company where a credit card and a couple clicks on Google or a few other places can get you to the same capabilities that a large company might take two years and a couple million dollars and people to put together. So, it really depends on where you’re starting from and where you’re trying to go. But I think to just say a blanket, across the board, No you should always go to a service, or you should always go on-prem, it all comes back to what’s that problem you’re trying to solve? What’s the data that you have? Who are you trying to empower? And, how unique are you and how unique is the data that you’re using?

The other thing that’s getting fuzzy is like I might say I’m going to go build this myself, but I might build it in Amazon using some of Amazon’s tools, or I might build it using tools on other cloud service providers.

J. Maguire:

So, even defining what’s in-house and what’s not in-house is fuzzy, you’re saying.

A Germanow:

For sure.

S. Sangani:

James, can I take the bait?

J. Maguire:

Please. Please, yeah.

S. Sangani:

I think you would be, you meaning anyone that is considering building an in-house data solution where they think they’re truly unique. You really ought to be in the .001 percent. I mean, if you are trying to understand how people use your website and how people are using your online software, and you don’t use New Relic or something like it, you would be insane. And if you tried to go build visualization and you think you’re the only person to do this and you don’t use Tableau, you’d be unlikely to be correct. And so I do think often in the world of technology if you have the hammer, i.e., I can code, then the nail of building something seems attractive and reasonable. There has been extraordinarily large amount of venture capital and  customer oriented development investments made in analytics over the past decade, so in-house seems like a very unlikely place to appropriately invest your IT dollars.

J. Maguire:

Thank you, I wanted to hear someone say that, but you put it really well so I appreciate that. But you never know, some people will still want to do that.

S. Sangani:

They will, absolutely. Yes they will.

J. Maguire:

What about… pointing to one particular data analytics technology that you would say, hey, keep your eye on this. There’s all these moving parts out there, all these things that are emerging, but this is the thing to really keep an eye on. What would that thing be?

At the same time, let me see if I can complicate that question beyond all reason. If we’re talking about what technology matters, what about usability versus feature set? I just, that’s a second question. Ignore that. But what is that one key technology. Dan, what do you point to and say, ‘keep your eye on this, that’s the big deal.”?

D. Kogan:

I mean, okay… Super layup. This is showing nothing interesting, but really, Spark has been an open source technology that’s now finding a ton of different use cases in the analytics space. There’s just so much going around and around that project. That is an interesting one, for sure.

J. Maguire:

It feels like the hype around Spark is overcoming the hype around Hadoop. I don’t know if you’ve sensed that.

D. Kogan:

Yeah, well it is in some ways. But also, they’re not competing technology, necessarily. You can work with the data and process data in Hadoop through Spark and Spark SQL and that engine, so in many ways they can be complimentary.

I think one we’re keep an eye on at Tableau would be doing more with this Presto, which is another SQL over Hadoop engine. We’ve worked with Hive and Impala and Hawk and many of the SQL over Hadoop engines. Presto is another one that we’re looking at optimistically. I’ve been having good performance. We’ve actually been really impressed in general with how that world has evolved and Impala is a good example of that really providing good interactive analytics use cases.

Not to throw too much of a nod over to my co-presenter Satyen, but we really like what his company Alation is doing and companies like that that help in terms of establishing the data catalog, and we’ll call it sort of the metadata management and the source of truth across and organization without having to do it in a really hard physical model that has historically failed companies in the past. So, that maps really well to a world where, again, as people source more and more data from a number of different sources, and it’s not all physically living in the same environment, how do you establish some consistencies and some truths across that data in those types of environments. And so their software is a really interesting one for that model.

A Germanow: I agree with that. Not to make Satyen’s head too big, but the notion of data dictionaries and just understanding what exists and what’s behind it. One of things that I’ve told people over and over is that we don’t need more data scientists, we need more data janitors, because one of the things that you see culturally within organizations mistrust of data. There’s this notion that even if we have the data, even if it’s knowable, even if it’s 90% right, well, some executive got burned three years ago because it wasn’t right, or it’s not exactly right…

J. Maguire:

You’re talking about with the scientist/janitor analogy, saying prepare and clean up the data before entering it?

A Germanow:

Yeah. When you have lots of noise the process of preparing the data and making it clean and analyzable is often where lots of people spend a lot of their time. So, anything that you can do that accelerates that action. Anything that you can do that brings clarity to what it is that you have is super important because one of the things that I think is kind of interesting is… the solutions that I think are interesting aren’t necessarily the underlying infrastructure pieces, those are definitely fascinating, but the solutions that help people do their jobs that are outside of the data analyst. So, the sales and marketing teams that are empowered by having all of the data that they need to know about an account or a contact. The software teams that have all of the data that they need. The finance teams. There’s a lot of different teams. So, the technologies that I think are fascinating are the ones that empower those professions in sort of new and interesting ways. That’s something that I’ve been spending some time looking at in terms of how do you kind of hide the big data-ness of the technology and yet empower users to go answer the questions that they need in ways where they feel confident about it.

J. Maguire:

Satyen, what do you think in terms of… can you clarify why you’re getting the kudos from your fellow panelists?

S. Sangani:

Yeah, the check’s in the mail.

J. Maguire:

But also, explain exactly what they mean so that’s really clear.

S. Sangani:

So, I think it gets back to this notion of curiosity very critically. So, it’s an elective exercise, and I know that sounds a little crazy, but you don’t have to think and you don’t have to ask questions and you don’t have to do things better. Those are things that you could do if you had time and you could do if you had some information. But, to get people to sort of challenge the status quo and to learn and to use those learnings to make better decisions is a tough thing to do because you’re trying to change the way in which people work in the way that they do and think in the way that they think. Now, if it’s too hard to use a given technology, so when you say usability versus features, I go completely on the side of usability because if you can’t use the thing you’re just not going to ask the questions. It’s going to be too hard. You’re going to go do what you’re going to do based upon gut instinct, right? As a secondary idea, if you don’t trust the data that you’re seeing, and that’s part of the problem that we solve, and we also help you find data, but broadly speaking we provide this catalog. Like LinkedIn is a catalog. Alation basically helps inventory the data inside of a company so that you can find the data you’re looking for, whether it lives in a tool like Tableau or a database like Hadoop. The basic premise is if you can’t find the information fast enough, just like Google helps you find answers on the web or Yelp helps you find information about local businesses. If you cannot find the data and then trust it, you won’t use it. So, there are real challenges with information, broadly. To get information to be used it’s got to be easily accessible, it’s got to be easily understandable and it’s got to be easily trusted. Absent any of those three things people will just do the gut instinct thing because you know what you know and this other thing just doesn’t seem accessible, trustworthy enough or useful enough. So, this is a psychology problem as much as it is a technology problem. I think if you want to get people to use big data and get people to use analytics you gotta make it easy for them and you gotta make them trust what they’re looking at.

J. Maguire:

Right. One last question. Kind of the crystal ball question. If you look three to five years out, where is data analytics going? And specifically, what are those coming changes going to mean to a business hoping to upgrade or get on board with their solution? How does that influence the shopping process now, thinking three to five years ahead? Dan, where is data analytics going?

D. Kogan:

Yeah. I mean to me it is still that we are just scratching the surface of the people who have access to the tools that can give them answers to their questions, both in number of companies and within organizations. So, I think that democratization or that access to data is just going to continue to grow. You’ll see… I think big data is a term that will go away more and it will just be ‘data’, and the expectation is you’ll work with all these different types of data and it’s living in different back ends and the middle is getting stitched together more virtually.

J. Maguire:

What do you mean by the middle in that?

D. Kogan:

You know, it’s the concept of an enterprise data warehouse for 20+ years where all of your data coming from all of the different systems that you deal with sits in one place where it can be cleaned and managed and governed and controlled, then the user giving out to the end users. That type of a model is just proving itself not to be scalable and not to allow people to answer their own questions, and certainly to do it quickly enough. The reason you have that was still so that you have the data you see is trusted, and you have trust in what you’e analyzing. So, the theory is right. It’s just the implementation of and the practice of how you get to trust the data and empowering everybody with questions in your business to answer those in a trusted fashion. That’s what’s changing.

I think we’re seeing moves away from, to Satyen’s point, if you build for ease of use versus completeness of features, necessarily, the latter may have low adoption if ease of you isn’t there, so what is the value of the technology, ultimately, if people aren’t adopting it? So, we’re having to change the way, both in terms of the technology we build, but also with people and with process and mindset, of how you solve this problem you’ve been trying to solve but in a completely different fashion. Using a different test set for it and managing the data source as really the currency rather than the in/output itself, and empowering people in business to find those answers for themselves. That’s where I see it: just continuing to go down that journey.

J. Maguire:

Abner, what do you see looking three to five years out? What is this conversation going to be about data analytics in the year, say, 2021 or so?

A Germanow:

I would agree with a lot of what Dan said. The data source are, I think, getting better. The collection mechanisms, the cost of collection, the cost of storage will continue to fall. But the way that the systems actually help people do their jobs, whether that’s a data analyst or somebody who simply doesn’t care, like, they don’t care what the data is behind it, they just want to know what they should do next or what they should be paying attention to. So, the ability for these systems to tell a particular job or employee or individual, ‘Here’s what I think you should be paying attention to,’ to kind of create that first spark so that person can then go ask the next ten questions. Today, too often I think analytics vendors give people blank sheets of paper and say, ‘Look at what you can go do with this!’. I think that will turn over time to saying, ‘Here’s what we think you should be paying attention to. Now go.”.

J. Maguire:

So, this is going to be more prescriptive?

A Germanow:

Yeah.

J. Maguire:

So Satyen, what do you see looking several years out for data analytics? And how would that inform a buying decision now?

S. Sangani:

I think that you start with the World Economic Forum. I think one of the interesting themes when I went earlier this year was this notion that machines are going to take over human jobs and that AI and machine programs are going to rule the world. I think that there is some real power that comes with the ability to understand data. And I think that power allows you to make decisions. It allows you to sort of articulate yourself in a way that will make you much more of a differentiated commodity in the workforce.

I think what that means for all of us as tools vendors is that what we will see is that really the tools are going to teach people how to use this stuff. So, I think that could come in examples where we give people suggestions in their roles about how to make decisions and how to think about decisions, in the way that Abner articulated just now. I think that will mean that we’re going to have to teach people about what is a good conclusion and what is a bad conclusion, and what you can trust and what you can’t trust. When you’re building a chart, how to build that chart, and what is the best way to exhibit it. That’s one of the great powers of Tableau.

I think that there is a lot to do to almost enhance the skills of people. I think what you’ll find is that the compute technology which has hereto been the very interesting thing, the sort of last mile of processing a query or processing a transaction, whether that’s Hadoop or Spark or something else, will be less important than just getting a human to be able to start figuring out how to answer and question and how to ask a question.

J. Maguire:

Well, I guess that would be exactly my question. You talk about how the solutions might provide that and it seems like some of it would have to be a human.

S. Sangani:

Oh, for sure.

A Germanow:

So, think of it this way: Today, if I’m a professional whatever, how do I prove success in my job? How do I prove that the little change that I made had an impact that looked like x, y, or z? Historically, that might have been me sort of BSing my way through a meeting or somehow posturing or something. You have this culture problem today in some places where you’re fairly data driven where people are empowered to say, ‘Here’s the impact that I have had on my job.’ And they can use data to sort of prove that. That is incredibly empowering.That’s both culturally important, and so, you know, when we look at, for example, successful software organizations, the more they make decisions based on data versus speculation by running experiments and going through those actions, you allow people on the team to prove things in their own work life. So, career success starts to come through in terms of, ‘Here are ways that I can prove that I had an impact on the business.’ The more you make those things data driven, the better off you are.

Now, that’s a big cultural shift that takes awhile. It comes through tools like Tableau and others, because today there have been many people who have been successful by not using data. I had the CIO of a major automotive company come to me and say, ‘How do I get my staff to be more data driven?’ The answer that we came to is that he has data nerds on his staff, they’re just ten levels down. So, what he started doing was bringing the data nerds to his executive staff meetings. Two things happened: The first thing was that the data nerd said, ‘Oh, we have that data. I can help you make that decision. And then the second thing was that the other people in the meeting stopped BSing because they knew the data nerd was there. So, just that cultural shift is crazy important. By the way, his instruction to the data nerds was, ‘Just sit there. Don’t say anything.’

J. Maguire:

He was to be the threat, so to speak.

A Germanow:

Everybody knows that that person has the data. But that transition that I think we take over the next three to five years is that data nerd is not just the individual in the corner, but those skills start to flow out broader across the team.

J. Maguire:

Yeah. That’s good. I think we should probably wrap up. I think we’re good if you either wanted to add one more thing, Satyen, or are we all good to go? I think we’re wrapped up unless you wanted to say like one last thing or anything? No? We’re good, okay. Excellent. I appreciate the expertise of the three of you and I’ll send you the link. We can all tweet about it. It was fantastic. I surely learned some things. Thank you very much.