<img height="1" width="1" style="display:none" src="https://www.facebook.com/tr?id=2854636358152850&amp;ev=PageView&amp;noscript=1">
43 min read

SBA 247: Haystack and Metadata Fundamentals with John Petze

By Phil Zito on Apr 26, 2021 6:00:00 AM

Topics: Podcasts

John Petze has been a pillar in our industry for a long time. In this week's episode, we discuss the topic of data and metadata and explore the emergence, growth, and adoption of Haystack as a semantic model throughout our industry.

Click here to download or listen to this episode now.

Resources mentioned in this episode

Haystack Connect FREE Online Conference

Project Haystack


 
Subscribe via iTunes


 Subscribe via Stitcher

Show notes

Phil Zito 0:00
This is the smart buildings Academy podcast with Phil Zito Episode 247. Hey folks, Phil Zito here and welcome to Episode 247 of the smart buildings Academy podcast. And in this episode, I'm going to be talking with john petzi. of the both while he's with sky foundry, but he's also with haystack. So he's got kind of a double whammy here, but he's gonna be speaking on behalf of haystack. And what I'm super excited about with bringing him into this is not only does he understand the data side, and the metadata, and all of that, and the importance of that, but then he's also helping to lead a analytics organization, a software analytics organization. So he understands the ramifications of implement implementation of metadata. So we're gonna see kind of both sides come together in his expertise. As he communicates with us across this podcast, I encourage you to go to podcast at smart buildings academy.com Ford slash 247. Once again, that is podcast at smart buildings academy.com for slash 247, it's there that you will find the show notes, you'll find the recording, you'll also find a link to sign up for haystack Connect, which is completely free, I encourage you if you've been trying to learn about haystack, at least sign up and go to the intro to haystack, and learn about what it is and what it means. You know, as john says in the episode, and as I like to say, you always should be learning and networks were really, really important to us. And those of us who learn networks, we've seen our careers blossom data is also really important. And for those of you who focus in and learn data, you're going to see your career blossom as well. So I encourage you, I really encourage you to go and consider taking a look at haystack Connect. Alright, with that being said, let's dive into the interview. Alright, folks, we're here we're talking with john and we are going to dive into a series of questions. I've been really excited about this. Because when john and i talked in the pre call we we had a really good conversation around kind of the history of data. What probably longer than this podcast interview will go. So let's dive in to First off, everyone's talking about data, whether it's digital twins, whether it's data modeling, whether it's analytics, but what exactly is data in layman's terms?

John Petze 2:40
Yeah. So I think that's really interesting point data is now fundamental to everything we do in buildings, right? And you hear it in all of these conversations. And what we're talking about is the information, the values that the different devices and systems and buildings provide to us and temperature sensor. That's data, right? The status of a relay, the position of a valve, the position of a damper, that's what we're talking about. We're talking about operational data in the built environment.

Phil Zito 3:11
Why do you think that people, and maybe I'm wrong in the US, but it feels like there's either two schools and it doesn't feel like they're meeting in the middle, is you've got the school that is very it centric, and they're completely comfortable with data. And they feel like, Oh, it's no big deal. And they've got the facility manager, maybe they're 60 years old, they've been working on pneumatics. And they transitioned to DDC begrudgingly. And they're like data, they hear that term, and it just makes their skin crawl. Yeah. Why do you think that is?

John Petze 3:43
I think it's because it comes out of the background of where they've been working. You know, and we'll talk more about this. But if you've been working with any manufacturers be as right and running it, whatever the brand is, in think about data you thought about, I have a graphic, I feed a picture, I know what my systems are doing. That's what I care about. I don't care about someone it terminology, I care about that. So that was the world. The topics of data are coming out because we're trying to do more things with the information and the data. And then you get into the concepts related to data. How do we interchange data? How do we describe data? Right? How do we consume it? How do we map it, change it, etc. And when you were just running the building equipment, I shouldn't say just when you're in the building equipment, that wasn't your focus, the focus was the equipment the data was away you understood what the chiller the boiler, the or handle was doing. Alright.

Phil Zito 4:34
So then, and I want you to kind of frame this in the mind of people reoccupying, post COVID with the importance of IQ, the importance of touchless experience the importance of occupant tracking and notification. Why does data now in my opinion, more than ever become important to buildings and more so Building operators and I would argue building tenants as well.

John Petze 5:03
Yeah. Yeah, we would agree. We actually say now more than ever, right? You need to be able to work with the data, you manage the data use it. And I think bursaries in I guess I'll give an analogy. There were things you could do with data that were optional to you, as a building owner, hey, if use data effectively, you can reduce your energy costs. Yeah, but you can join, but you know, I'm busy or I don't, I'm not comfortable with return or whatever, you know, the return on that whatever, right. It's not optional to make sure you have a safe environment now, and your tenant, your occupants are more keenly aware of that. Right? And, you know, it used to be well, you know, yeah, your temperature and your space, it can't quite hit 74. Phil, sorry. That's the limits of the building on the south side, you know, and people grumbled with it. But now it's like, What do you mean, you can't deliver fresh air? What do you mean, you can't track? You know, proxies for air quality? Right, change, you know, it changes for our co2 levels. VO C's, what do you mean, you can't do that? Now? It's not really well, everything's optional, but it's less optional than the optional benefits I might get by trying to learn to use data. I think that's it. Part of it, you know, people recording this. I don't get the webcast. I mean, they're recording their data, because they may have to prove to their occupants, tenants, right? leases are, are have, for a number of years embrace comfort factors probably weren't going into maybe they went into co2. Maybe they didn't. But now, yeah, I'll come back in the building. I'll renew the lease, but you're going to confirm report that you're achieving these required levels. Okay,

Phil Zito 6:45
so what I just heard you say, and I want to expand on this little bit, but what I just heard you say is that as people reoccupy and renew leases, there potentially could be a compliance aspect to it around IQ, around visualization of IQ reporting of IQ. So that's why I just I don't want to put words in, you

John Petze 7:09
know, that, that that is what I'm saying. And that might be private compliance. May is the tenant with you as the, you know, the landlord, we're going to, or it could move into regulatory environment. I mean, you know, you think about ASHRAE has put out guidelines, well, what the guidelines turn into, they turn into requirements, right, over time, you know, state legislatures or local building codes, adopt them over time. And I think with the focus on this, and the concern, these are some standards that might get adopted more quickly than others that were around, you know, energy efficiency and things like that. So, you know, that, but I believe there's a factor there, it's not every building, but that this is a factor that I think is highlighting the importance of data for people who may not have even given an a consideration. Interesting.

Phil Zito 8:00
Yeah. And, and something I don't expect you to answer this. But one thing that pops in my head when I hear about IQ, and I look at the guidelines from ASHRAE, increased ventilation from outside increased air changes, running these systems more. And then we also look on the flip side that societal society wise, we're focusing on energy we're focusing on net zero and two seem diametrically opposed.

John Petze 8:31
Yeah. I wouldn't say diametrically opposed. I think there has to be a balance right. You know, let's make an extreme case. Phil, I can save you 99% of the energy and your build time. Oh, is it? Yeah, right. Yeah. The the energy is there to serve a purpose? Well, the purpose is a balance, right? We used to say, okay, with 74 degrees, that's all you get. But now we're saying there's other things we need to balance. And yes, it may have an impact on energy. And what should come out of that is okay, how do we help offset that? Well, now let's attend to the efficiency measures that will relate it to things not operating properly? Right. Not giving people less, giving them what's needed. But making sure we're doing it efficiently, which, you know, I believe in from what we see in our world, the analytics is still a huge opportunity, buildings run so inefficiently. And maybe this is a new impetus to make sure people look at the operation performance and maintenance of their systems.

Phil Zito 9:34
Yeah, especially if they're stretching them to the limit to meet these IQ standards, because you're gonna have things that were not designed for that capacity. So you mentioned working with data when I asked what data Yeah, why data is important. What does that mean to work with data because I used to sell analytics back in a previous life. And I would say, data and analytics and we're going to work with the data and you're gonna work with the data and people would start to glaze over, they would say over dinner. So you want me to be a data scientist? And I'm like, No, I don't want you to be a data scientist, I just want you to be able to work with the data. What does that mean?

John Petze 10:14
Well work with the data is part of the process to get to the end value. That is what most of the people who run buildings care about. They don't they don't want to work with the data, they want to know. My KPIs. What am I operating KPIs? And how are they trending? Well, to get to that point, someone had to work with the data. It doesn't have to be them. And I think that's a good point to bring out is that many of the people who own or operate buildings, they don't want to, and they shouldn't be forced to say you're going to work with the data. That's not necessarily true. But they want the end output, they want the information they want, as we say the things that matter to me the things that matter. Okay, your key KPIs that you we've agreed you want to track, you know, consumption per square foot consumption per square foot per degree day per occupant per hour is about a trending in the wrong direction, right? Your energy use it profile is getting worse, not better. That's what you care about, how did I get there? Well, I had to get meter data, and I had to get scheduled data. And I had to bring those things together. And I had to run some math on them and I had, but the end result is, here's information for you about the things you care about, right? The work what data is for, and I won't anyway, while they're data science, roles to play, the vast majority of what we're talking about with analytics and buildings, you can categorize it as data science, if you want to use that new buzzword. This is just this is basic stuff, right? You know, it this is not deep, deep science, although you can apply deep science here and more companies are to get at, you know, to do more things with the data to do more predictive models with the data. But on this subject of Why do you care about data, I gotta throw I'm gonna throw out a thing we like to like to say sometimes, especially when we get to do it in person when you read people's reactions. But, Phil, do you know that the data produced by the equipment systems is more valuable than the equipment systems themselves? And people think for a second Oh, well pick a VAP box controller, what's that cost? 300 400 $500? Run it run it wrong for a year. What did that cost? Right? Well, it certainly can't be true for a chiller. That's hundreds of 1000s of dollars. Okay, run it wrong for you. Right, you know, the data is more valuable, but because it's intangible to the average person, it's kind of hard to make that connection. Especially if you don't touch the data, you just look at, yeah, I got a picture of the air handler, everything's fine. Right?

Phil Zito 12:49
Yeah, I would expand on that to say, not just from an operational perspective, but from a procurement perspective. Because I would see all the time, people convinced that they needed to buy another chiller, or that they needed to expand their plant capacity. But then you go analyze things, and you're like, well, you're heating and cooling at the exact same time. You've got your outside dampers 100% open and your cooled chilled water valves 100%. On a mixed air unit that's satisfied at co2 return is satisfied. Why are you doing this? And so I think they missed that as well. We focus so much on the opposite side that there's a cap x return as well.

John Petze 13:33
That's, that's a great point there is. And we've seen numerous cases of that where through the use of data, the final consumption of IT people learn about their systems, and how they can be optimized without those capital spends, right. Oftentimes, I would say that's counterintuitive for people, that guy who's running the chiller full bore in the bowels for war probably thinks it needs to be that way. because no one's ever taken the time to do a more detailed analysis of it. Right? That's very common common thing. We saw that specifically your exact example, about an organization sizable organization, large campus that thought they needed to expand their chiller plant the central plan. And they got major pushback from the financial part of the organization, not because they knew better about energy because they said Phil, you're not getting the money. What else can we do? And they went in the way and they thought we don't need to expand the chiller plant. We really do have capacity here. But we have been using it wisely. We haven't been attending to other problems, okay. here that are wasting energy. So it definitely has a play there. I'll give you another example of how it affects capital purchases is how about you know, and a lot of people I think we'll get into you know, data and analytics often is equated to fault detection. Hmm. That's one use of analysts. Next, right here, you're tracking KPIs and trends that aren't a fault that just trends. That's another. But here's another recording and being able to report an analyze based on the number of service calls per HVC unit per vendor. Why do we have more service calls on vendor x than vendor? One? We didn't know that, hey, guess what we're gonna do on the next bid? Or the next procurement? Or how about energy per vendor? Right? Or per building design star, you know, there's people have, you know, even in multi site retail, they might have three or four designs, right, which one really performs better. And a good example we have that was a customer. And they analyze their building stock, they were very early on, they weren't even connecting live, they were just taking past energy data. And they found out some of the older buildings built in the 70s. And 80s outperformed the new ones that were supposed to be more efficient, right? They were they were shocked by that. Analysis brought them to that understanding, which then you build off of Where should we go, right? Is it the new buildings don't work, right? Or, you know, this is this design was flawed, or we're just not, you know, we've deferred the maintenance. So that's why they're not working, right.

Phil Zito 16:14
So or you give people more control, and now they're starting to run the system efficiently.

John Petze 16:21
Yes,

Phil Zito 16:21
that's one of the downsides of smart buildings is, you know, everyone touts this whole edge building. And I see it used as a model for everything. I don't really like that. Because when you dig into it, you realize that it was really financed as kind of a marquee showcase for a lot of people, and you couldn't really fund it in real life. But if you dig into it, there's articles about the guy who manages the edge talking about data overwhelm, and the best one was that people would go and they figured out that if they booked a parking space, it would automatically reserve a room, but there weren't enough rooms. So you'd have people booking 510 parking spaces, so that they were guaranteed a room when they would come in and it would occupied these rooms during the working period. I'm sure they fixed that. But it was a really interesting, I got to find that articles. But that's key points. Good intentions.

John Petze 17:17
Yeah. But they may have fixed it. How did they fix it? The data showed them the problem. And that's what we're talking about the data giving you the awareness of what is happening. We had another example of a building that was designed for net zero operation, right? Brilliant design all of that a couple. It didn't work that way out of the box. Why? Well, through analysis, they understood the interaction of the control sequences with the building, and they did achieve their goal, but not by but not without being able to have tools to look at the data and understand the dynamics of these highly complex systems we call buildings.

Phil Zito 17:56
Yeah, that zero is really interesting to me. That's a whole nother topic. But

John Petze 18:00
yes, yes, yeah. Yeah,

Phil Zito 18:03
I will say one thing that I think so when I was selling analytics, it was 2012. And one of the challenges I ran into was it was very early from a technology stack. And I, one of the things I love about our industry, but I hate about our industry is, you know, I suggested the other day to a guy, hey, go hire some test and balance people they know, because he needs B is techs and I said they already work with ba s, they understand h fac. And he's like, yeah, I hired a test and balance guy back in the 80s. And he wasn't any good. So I'll never hire test and balance ever again. And I'm like, for real Dude, this the 80s? It's like that with analytics. I went and did this analytics solution in 2010. Yeah. And it cost a lot. It was really difficult. There was no metadata, which we'll get to in just a second. And so is very expensive. I'll never do analytics again. Yeah.

John Petze 18:59
This is an odd human dynamic. Right. You know, what else has changed since the 80s? Are you still wearing your bell bottoms? or? Yeah, I mean, so why wouldn't you expect that these technologies think about what was the phone you were carrying? And what did it do in 2012 versus today? Well, we can accept that these other things may have advanced that people may have advanced. I hope I know something more than I did. 10 years ago. Right? I yeah. I don't know why people fall into that. I think it's an easy excuse. You know, that that gets used by I think all human beings in different ways. But we do see that in the buildings. Oh, yeah. We try to you didn't have these features. Yeah, 10 years ago. Yeah. We're taking a fresh look right. You know, and that just holds people back from accomplishing things that in other parts of their life they do, you still watch in a tube TV, you know, or you get a flat screen or you get an old lead. I mean, TVs Terrible, you know, the resolutions. You're watching a tube, you know, playing VHS tapes. I mean, I don't

Phil Zito 20:06
get it. It's kind of funny when you look at people's consumer habits, yeah. don't match their commercial happiness.

John Petze 20:13
Yeah, exactly our industry, right?

Phil Zito 20:16
Yeah, it's so let's speak about metadata and haystacks. I'm a user of haystack. I've, without biasing this interview. I've personally enjoyed it. I haven't used brick and I haven't used Google's new schema that they have. So I can't judge those. Yeah. But since haystacks so easily built into Niagara, and Niagara has got such great market share, I find us using haystack quite a bit in our instruction to our students. Yeah. But what is it? How did it come about? And how is it related to data and analytics?

John Petze 20:54
Yeah. And I think that's where we should start, forget about which standard we're talking about. Why do we need to apply metadata, semantics or tagging? Those three words are basically mean the same thing when we're talking about this? Why do we care about why do we need it? And it's a great question. Because if people haven't encountered the reasons, they're not obvious, right? And the reasons have to do with our increased desire to be able to use data from different devices and systems together to learn things, right to combine operational data, status of equipment, against energy data, so you can see how they correlate. Those are two different systems, you've got to bring it together. And then you can see interesting correlations, right? Like, we're slamming on all of the reheats at once in the morning, causing electric big, you got to happen. So when you're going to combine data together, it has to have meaning that can be interpreted. And I think we can start at human interpretation. Right. Your name is Phil Zito. What does that all alone? Tell me? I don't know your age. I don't know where you live. I don't know what your job is. I don't know what kind of car you drive all those things? What are those? Those are attributes, that different processes, you know, whether you want to get medical insurance, housing, those organizations need those attributes, right? The tags, the tags on you, right? They tag your age as a tag, right? Your address is a tag, if you will, it's a number of tags. But that's needed, because the insurance company can't make a decision on you just by talking with you. There has to be some data record about you that they can utilize. Well, that's what we're talking about. When we talk about tagging, we're talking about adding attributes, but the you know, what's called tagging or metadata. And I think the next step of that is why is it hard for people to initially grasp this? And my, you know, how I came to understand what this barrier is, is they didn't have to deal with it. When you had that VA s installed by company x. They set the data up, you've got a screen, it's an air handler, you understand it? You know what it is? And it's got data on the screen? Well guess what behind that, whether it's based on a standard or not, is a data schema with data attributes. That's the only way that graphic could work. So data metadata, it has been a part of everything we've done in VA s from the early days, right? It might have been incredibly simple. It was proprietary, it was hidden, I didn't have to deal with it. Or I didn't know I was still living with it. Think about, you know, now I'm a technician, I'm setting up a VA s, I have to define that analog input, a dialog comes up. I have to fill out the fields in the dialog. Do I sit there and go? Why do they make me fill this out? Right? No, it's needed. So you can use the sensor in a control sequence. That's metadata, right? But now what we're talking about with haystack and with the whole concept of data semantics for the built environment is

John Petze 24:06
another way to describe metadata, its meaning it's how we describe meaning of things. What is the meaning of that word? What is the meaning of that data? So semantics, metadata and tagging kind of get used interchangeably? I think that confuses people. I prefer to use tagging, but if I get into a conversation with somebody more technical, more software oriented, you know, I let's talk about metadata and semantic modeling. Right? So but they, you know, depending on your level of conversation, they all pretty much mean the same thing. Alright. Why is this now important in our industry is we want to use the data from those different systems for other purposes, then their native views in the VA s one the graphic, and we want to do that cost effectively. Right? We want to do that without a lot of manual effort. If I said to you go in manually get The current temperatures on the air handler out of your Bs and write them down on a piece of paper and bring them to me and hand them to me. So I can put them in a spreadsheet to do analysis, right? You could do that. But if we want to scale that, that's not gonna work out too well, right? So you say, hey, john, I can do this much better, we'll connect and we'll get the system. So we connect, and then we'll get the data. The problem is you're getting the data without any modeling, tagging that describes what it is. So I'd say, hey, great, Phil, you got me the data come over and tell me, what's this one mean? What sensor? is this? on? What air handler? You know, is it the mixture of the supplier all these basic things I'd need to know. And so Phil would come back over drive Hold on. That's not too efficient. We can't scale that either. So you don't want to waste your time doing this. So you say, you know what, I'm gonna define the things john keeps asking for. So when he gets the data, it has it. And I'm happy and you're happy, except you created a proprietary one off model of attributes, attacks. They've helped us, you and me, didn't help anybody else didn't help the next guy and the next guy and the next software application, etc. So what haystack is getting to that question is a community open source effort to agree on the attributes and how we're going to express them, so that if I support haystack, you get the data you need in your application in a way that can be interpreted, right? And I like to use this analogy. Look at HTML for webpages, before our call, I went to your webpage, I was able to read everything on there. I didn't have to get one of my software engineers to say hey, I want to go read Phil's webpage would you get with his software engineer and find out how they model the text so we can interpret it, right? That was long ago solve the web wouldn't exist without it HTML, open standards body came together and say this is how we're going to tag or describe data in a file so that a web browser can read it and display the same thing to everybody. That's what haystack is for the operational data coming out. That's the problem it solves. It works differently as we're not talking text and pictures and column formats. And all that we're talking about, what are the essential attributes or meaning of this data? This is a zone temp air sensor expressed in degrees Fahrenheit. Yep. I just told you six tags. Not so mysterious right now. That's what you know, that's what you finally decided to get me off your back. Okay. Every time I give john data, I'm gonna say, john, zone temp, air sensor degrees Fahrenheit. In fact, I'll put it on columns in an Excel spreadsheet. And if I presented it that way to you would think nothing of its Yeah, basic, I need to know this. Well, one of the cool things about haystack is it's human readable and machine readable. A lot of times people will enter the tags right in an Excel spreadsheet, because that's the tool to have at hand. And they are at the stage that the system or device they're working with, doesn't express that information in any digital way. So they say I can read the value, but I'm going to have to manually say what this is. And what this is. Remember what it used to be. I know you do. It was descriptions. But what's the problem with descriptions? Right? It's like your name. Let's go back to your name example. Okay, I've asked you all these attributes. And so instead of creating attributes, I'm just going to build it into Phil Zito 40 years old, this that the other thing drives this lives. And your name now is 1000 characters long.

John Petze 28:59
Not gonna work. That's what we're talking about with tagging and haystack being an open source agreement on how we're going to tag. The first thing we agreed on is the vocabulary, right units, zone, temp air sensor, those things which are easy for people to absorb it I will back to the BS thing. You were entering similar stuff to set that point up on the BS. You didn't know what no one was talking about it being tagging. It didn't seem like a problem. It was just like, I've got to do this right.

Phil Zito 29:32
I will say one thing that has been interesting to me about this whole tagging journey has been you get pushback from people about data about analytics about metadata. But then as you analyze their organizations and you see that they're trying to operate a campus with limited staff, and every building has a different name for a point. That's like you do realize, like when I talk to owner operators and they're like, what can I do? That's pretty low effort. As far as not like having to physically replace things that will increase my efficiency. I'm like, you can standardize your naming and metadata because then you can train someone once on the air handler, right once on the VA v box, and it's always gonna look the same. Like, I don't I still don't get why people don't grasp that. It seems so freakin obvious. Yeah. It's like if if I named something door, and then I named it gateway. Yeah. And then I named it portal. And I'm like, trying to communicate. Where's the portal? Personally, I don't know what a portal is. There's the gateway. I don't know. Where's the door? Oh, it's over there.

John Petze 30:51
Yeah, right. And I want to separate those two things. But as you said, standardized, standardized naming convention and tagging. There are two different things. I think that's really important. These people often misunderstand haystack is not a standard naming convention, it does not tell you what to name your points, that would be impossible. We do however, say you're going to benefit if you have a standard naming convention, because you're going to be able to automate the addition of tags, you're going to say when I use this character string as a TMP, dash number, that is supplier temp, air handler number. And if I do that everywhere, now a script can go and add all the necessary tag hastac tags, but they're two different things, just like your name, right? didn't capture your attributes, even a standard naming convention. You know, like a lot of companies do standard email conventions, Phil Zito, yeah, that didn't great. We have standardized email names, anybody can figure them out. It didn't do anything about all those other attributes. So standardized naming haystack benefits if people do it, haystack does not impose it. It is not a standard naming convention. It is a standardized method and vocabulary to add the other information to define the data. And its mean.

Phil Zito 32:06
Yeah, and I'm glad you pointed that out. Because oftentimes, when people have a campus, and they don't, and they're looking at applying haystack, or they're looking at applying metadata in general, it can seem challenging. But that's what I say, if you have standardized names, then you to your point can do like a Python script or something that goes and adds that. And then I still don't get why. I mean, I kind of understand but I don't at the same time, why OEMs are just not really dumping their dollars into a standardized metadata platform and process versus all the other stuff they're trying to develop. Because if they haven't been on the r&d side of things at a large company haven't seen how they develop things. If they had a standard data semantic model, it would make everything just so much easier. I think that's the reason why so many of these startups fail is because there's no common model to pull data out. Like you mentioned HTTP and HTML, HTML. Yeah. Imagine trying to build web apps without that.

John Petze 33:19
Yeah. Insane. Let's expand on this because I can relate to our experience in launching sky foundry, you know, back in 2009. And, you know, I'm kind of the domain expert, and Brian Frank, my partner, he's, he's the software guru, right. And we're going to do data analytics, and hey, john, he's educated me, we're going to need to deal with this semantic thing. Oh, yeah. Oh, wow. Okay, make sense? So we, john, go out and find the industry standard on this for the bill, buildings market? I go. Okay, I'll go take a look. Maybe so, you know, maybe some papers have been written, it didn't get widely adopted and dead end? Yeah, I

Phil Zito 33:59
know. It's amazing,

John Petze 34:00
dead end. And this is 2009. Right. And so, you know, hey, guess what, Brian there, there isn't one out there. You know, okay, well, I'm going to, I'm going to look to the lessons of the IT industry, the Semantic Web and concepts around how we do this, right? And will we have to do it. So we're going to implement a model, and we're going to draw it from the world of software. I implemented it. And now it's 2010. And we're selling and now it's coming around the end of the year. And we get down and talk about, you know, this is not a competitive advantage. This is actually the barrier to adoption. And so we talked to other people in industry, some people at the research labs and say, Hey, what do you think? And they're all like, yeah, this is a huge problem. We said, Let's get together and launch an open source effort. And a small number people said, Yes, that's what we need to do. And so we set up a server and we at Sky foundry turned over everything we'd developed right. If you know Brian, Frankie takes a very standardized approach. We submit it turned it all over open source under the academic free license. And we started trying to rally people around, hey, let's work together to solve this problem for the benefit of all right? And that in March 2011, just a month, over 10 years ago was when haystack was launched as an open source, community driven effort. And that's the other thing about this. We didn't develop it all and dump it on people and say, oh, but it's open source, right? We've been working with literally 1000s of people around the world, not all of them contribute as much as other to solve these problems, right? We had a really big, I gave you a good example, he had a really basic model for electric meters, and electric distribution panels. And somewhere in that community is basically, you know, you guys don't know crap about me, you have all of these attributes, and I'm in all of the, hey, Phil, come on in, help us expand the vocabulary model for electrical distribution, be as you know it, and put it out there for consensus, debate and approval until everybody basically fritters away, no more complaints that's in the standard. And that's what we've been doing. And you know, we got haystack Connect coming up. I know we're going to talk about later. But we've got two new working groups that they are tackling aquifer based thermal energy storage systems out of Europe, one of my areas of specialty. But there's a team of people that that's really important to them. And so they have been working on the tagging model for that. Right. And the other one is for vrF, verbal refrigeration flow systems, which as you know, a really those are huge, they have huge taken off more and more. Well, guess what one of the major manufacturers are going to propose a model for that.

Phil Zito 36:46
That's kind of cool.

John Petze 36:47
Yeah. It's very cool. And yeah, because it happened by me. That's the community working together to solve this massive challenge

Phil Zito 36:55
as this knee because I mean, up to this point, we've just had back net and its object model. And that wasn't a metadata model. That was an object model. Exactly. A lot of different. Mix that up because they're like, Yeah, but BACnet has devices, and it heads as objects, and it has an object model. So why do we need a metadata model? And I'm like, because that's like saying, you know, you've got a car, and it has doors, but you're not describing how many doors? You're not describing what color the car is not describing how many tires it has. Exactly, yeah, good, good.

John Petze 37:29
Now,

Phil Zito 37:30
don't realize that.

John Petze 37:31
No, they don't. And in fact, that's, that's another challenge this whole thing, because you'll bring out, you know, some of the things we've learned in the conversations and trying to educate people and project haystack hat back on here, is people say, you're talking about interoperable data, john, we've had interoperable data since BACnet. But whatever, pick a standard, I go, now, you didn't have interoperable data, you had integrated data, you could get it and you could do the extra work to integrate it by getting Phil over to explain what these crazy names mean. And help me decide what they are and what piece of equipment there is. And then if I do all that work, it all came together. That's not good enough. And my analogy there is, you remember when setting up a printer was like all day or get the IT staff in, right? Well, you go down to the store, you buy, you plug it in, and you know, 99.9 times out of 100, it just boom, it's working. How did they do that? standardized plug in play? Well guess what's one of the key element of that? standardized? Well, you know, here's who I am, what I am how I come in all those. That's the only reason we have plug and play. We could integrate printers 20 years ago, but it was painful. All right. So this is the same thing. And we're trying to come out of that pain, because we know we want to use the data from these different and growing amount of devices in buildings, right? These IoT devices, Wow, great. This is really valuable. Yeah, if I can combine it with the other information, right, then I get more value, the network effect.

Phil Zito 39:03
I also think it's one of the biggest barriers to scale in our organizations, because we can't really develop efficiency tools, because so many of these efficiency tools require standardized data to be able to act on them. So you know, using scripts to auto generate graphics, to auto generate programs. I mean, it's still boggles my mind that to this day, we still have custom programming, I don't get why we do that. There's an I know, people will argue there's an infinite amount of sequences, but if you look at the Pareto principle, there really is a finite amount of sequences for HVC system control. And the fact that we have to program them custom each time, or use templates and just I don't know it's just inefficient. I

John Petze 39:57
want to break this down a little bit more is I want to pick up your first example Because I think it's a very good clear way to show the value graphics, right? You know, I've been around in this industry, the BS industry since the first color graphics emerged. First time ever you get out of, you know, a line drawing of an air handler with lights flashing value on it right. And over time, my saying is those have become the tail that wags the dog, right, more development effort, more focus, and I would argue more, more engineering hours on projects went to drawing pretty graphics than to do in a really good control sequence. Right? And why was that right had no other way. And there's all the demand for sexiness, the graphics, but here's how tagging affects that. There are products on the market today that they can look at the tag points know what they are, what piece of equipment, and they automatically assemble those beautiful 3d shaded graphics that everybody wants to see with zero engineering effort. Okay, I been involved when some of them on that concept was presented to not unsophisticated building. And they actually did not believe that they were being told the truth because they were so ingrained, that it's an hour per graphic, right? Or whatever. 30 minutes, right, that goes away. Now, let's go to you mentioned about control sequences, we'll leave out the question of, are there a finite number? Let's, and I think this is one of the things that hurts our industry, people want perfection, black and white. If you can't do every possible, you know, control sequence out of the box, then you need total programmability from the No, right? Okay. You can write control sequences, have them there, they reference tags, as soon as you add equipment with tags, the control sequence, finds it and starts controlling it. That's happening today. If you have tagged data, then, you know, I think you get to a philosophical question about how far how close to zero, can you get on the need for programmability. But you can get well down that road. We've seen it happen with terminal units, they're configurable now. And air handler controllers are pretty much configurable. It's at the system levels, that you may need programmability. But I'd also say some of the products haven't exploited the fact that terminal units and air handlers can be configurable versus having to be programmable. Right. So, so programming on saralee,

Phil Zito 42:30
I'm going to make a statement that may be a little controversial, I think that the only way you're going to see that really happen is either a OEM is going to agree to take a massive cut on their branch structure. Or it's going to be a startup, because if an OEM goes and rolls out something that significantly reduces labor, either they're gonna have to retrain their sales force to go and sell integration or sell some higher value. Yeah. Or they're gonna have to get rid of all that staff because they're going to have greater efficiency.

John Petze 43:08
Yeah. Nobody likes a top line go down, right?

Phil Zito 43:11
Yeah, those are billion dollar branch structures, that you don't turn the tight. You don't turn an aircraft carrier? Yeah. Well, actually, aircraft carriers are pretty maneuverable. They say you can't turn it, maybe a cruise ship is so I, it'll be interesting to see where that development comes from. Yeah, whether an OEM is willing to bite that bullet.

John Petze 43:34
Yeah. Yeah. Well, that gets here. You know, I think you mentioned earlier, you know, where's the adoption? Or we're going to talk about where's the adoption with the major manufacturers, obviously, tritium has embraced tagging, and they and they include haystack among, you know, and that's the other thing, one of the things about haystack is it's an extensible mod methodology. in two ways, one working groups can come together and say, Okay, here's the improved model for electrical distribution. And here's the first ever model for you know, aquifer based thermal energy storage system at all or not. But the other thing is, you can use the methodology to create custom tags you need for your own purposes. And they're still readable by an external application, it would have to know what it wants to do. But it still comes out in the same defined methodology by the API and the data structure, right? So it's immediately readable. On the so you know, where do the different manufacturers stand on not only adopting it, but publicly embracing it? And those are two different things. And I'll give you an analogy. Yeah, I'm executive director project haystack and I draw a really hard line between that and my day job. That's my night job. And I know, there are people on our forums who work for major manufacturers in their product groups, but they register with their Gmail addresses. Okay. And I know there are companies using haystack that don't admit it don't advertise it. They want to keep it hidden. Right? It's unfortunate because I think it's holding back the whole industry, because it's not we're not seeing the network effect that people are really doing this. Yeah, we're not helping people understand, oh, this is this is just part of our lives today. Right. It's it's part of where everything's going to work with data. And it's, it doesn't help with the, the guy, the people who asked, Well, how widely deployed is this? So I can't, you know, I don't have the right to tell you that manufacturer x is using it. They just want to admit. Yeah, you know, and that's been an odd thing that I tracked back to the protocol wars. Right. Yeah. Right. Everybody went into a camp. And you know, at that time, you know, we are tritium and we fully support it back then and fully supported lawn, and it created some of the most amazing conversations, right? We're going to talk about BACnet tools, and then on the coffee break. Good, good, good. You're against that long work stuff. And I'm like, Well, I didn't say that. Well, you're doing all the BACnet. Yeah. And then I go the next main show our lawn tool. Good. You're against that back. That's no. What's warfare? Right.

Unknown Speaker 46:15
Yeah, sure. Whatever,

John Petze 46:16
I'll make you make a buying decision. Right? Yeah. Like my bus? Yeah, exactly. It is a great example of why you need metadata. Right. But anyways, is that you know, I think, you know, that pervades our industry for a decade or more, and it did end up as a war. And it was not productive. But I think manufacturers became very gun shy, there was some big manufacturers who went all in with one or the other and then had to switch or whatever, right? We names out. But I think it created a psychology people. Yeah, okay. We'll just watch this. It's pretty good. We'll use it but don't say anything yet. You know. And that's holding back the momentum of the industry of solving this in a universal way for everybody.

Phil Zito 47:04
Yeah, I think it's disappointing. And then having been at OEMs. In the past, knowing that, especially with something being open source, they'll often take it and then customize it onto their own. Yeah, rather than just following it in its purest form. And that just further exasperates, the issues, and it's like, what is that old saying of something's biting your foot to save your hand? I don't know. But that's what they're doing. They think they're protecting some sort of IP, when they're, I don't think organizations have realized that the product really isn't their product anymore. We're all pretty much on parity. As far as products go, it's really the ability for the tools. And then the higher level applications. Yeah,

John Petze 47:52
yeah, the value that can be derived from the data as more valuable equipment and the solution has far more value than the data. Right. But I think it goes up against the long term and green industrial models of a lot of the BS suppliers that they financially track based on selling things in cardboard boxes.

Phil Zito 48:13
Right? Absolutely. And it's tangible, how

John Petze 48:15
many cardboard boxes that we shipped this month. Right. You know, and and i think that that's part of it. Now, you mentioned will a startup solve this? You know, I think we've seen you know, I've been involved in a number of startups in the BS industry, you know, with differing degrees of success, you know, trading being a very different type of, you know, technology because it was sold to all of the manufacturers. But it's very hard to even if you have the best solution, it's very hard to break through to wide scale adoption. And the time in our industry is so slow. I don't, I believe the only way a startup can cause change will be by the threat they exert, not by the revenue model, but by the revenue amount, but the threat. And one of the OEMs are more, yeah, we got to do this. Right. We got to do it, right. It's, yeah, well, that may be the way they buy them. Yeah, that's in That's what I meant. We're gonna we're gonna grab this and make it part of our business or whatever. And so at the end, the industry did change, right. I mean, it's a very different industry today with so much of it using niagra base technology. Then, before that happened, right? It didn't solve every problem, but it's very different because they're getting competitive service competitive bid competitive install around Niagara based products, right.

Phil Zito 49:39
Let's switch gears because I don't want people to miss this part, which is haystack Connect. So it's at haystack connect.org. And it's coming up here. It's going to be may 4 to the sixth. I really liked that y'all did it kind of over the lunch break. Yeah, I feel that that was a very creative use of time. Because that's Kind of when everyone, you know, they wrapped up their morning meetings. Yeah, so that's a good time. There's several things that popped out to me here of like the haystack in practice and then the technical track. So as I look at the haystack in practice, and I look at the education market, that's the one that I am personally most excited about right now, because of all the money that is earmarked for IQ requirements, depending on who you read, it can be up to $100 billion earmarked through the K through 12, and higher education market for IQ and hv AC and just those kinds of improvements. And so let's talk about just haystack Connect, in general. And then let's talk a little bit about these different tracks in these different sessions. So take me through this.

John Petze 50:49
Alright, so if I told you 2011, we launched this Open Source Initiative, and we set up a server and we got people together, and we did it all, you know, email, and this and that. And we had a discussion forum when we got the effort roll. In 2013, we said, you know, what we get, we got to pull the community together, right? We got to have a technical meeting. And we created the first of that haystack connect 2013 Chattanooga, Tennessee. And you know, we launched the meeting, and we had over 150 people show up, right? And I would say that more than half of the people who showed up said, john, I don't know why I'm here. But you guys are doing something that sounds important that we ought to learn about. So I made the trip. Right. And that was that was the launch of what's it about? Why is it about? What's interesting about some of the keynote speakers, we had the Department of Energy, keynote speaker, high profile guy, and someone from Autodesk, who painted the picture in 2013. That all of the data, the design, the operational, the sensor will, it has to and it will merge together. And if you think about it, that's where this digital twin thing is going. Right? Okay. And so we created this event, and it's technical presentations of varying degrees of technical, right. So our business and applications like you identified, some are deep for software engineers. And he was one of the unique things about it, that was really cool. You know, every other conference, you may go to trilliums conference, or Johnsons, or whatever. Those are controlled by the company, the manufacturer, right? There's rules is, you know, rules on what's acceptable, and what's this, but it's also it's more than just rules. It's, it's defined by their culture, their singular culture. Here's an event that isn't controlled by any single manufacturer. It's where people working on open source come together to share and solve problems together, right? So there'll be a small table with those guys do an aquifer based thermal energy storage systems, he is there's only 10 people will get it right. And there'll be a huge table of people arguing about electric meters, right. And there'll be all these presentations. So it's technical presentations. And we made it biennial every other year, we didn't think we couldn't take on the load of trying to produce it every year. And you know that and but we felt that was frequent enough as the work continues. But this when we come together, so we did it 20 1315 1719. And for those of Ben, I know this probably some of your listeners, it's a fun event, because we're a little looser than in corporate environments. We have a band, we have all. But guess what happened? COVID happens?

Phil Zito 53:25
Yeah.

John Petze 53:25
So in an odd way, I think we've all seen negative and positive business effects out of when talking about the effect on humanity and our families. But we've seen some things are helped by this and some things are hurt. Oddly enough, going to a virtual format. This is purely virtual, right, has helped one thing. There were more people in the world who wanted to come and participate and contribute than could ever get authority and money to travel. We are over twice the last of we're already over twice the last attendance, right? Yeah, well over 500 people and we, you know, and we you've got past history shows 35% of the people sign up in the last 10 days. Right? Okay. Yep. So that's a benefit. The other reason it's a benefit is you know, we're allowing pre recorded and live so the guys in Singapore or Australia or Czech Republic, they can pre record and their content still gets delivered as part of a live continuous event, right? Lots of organizations have done have done this. And there's a great diversity like, yeah, it wasn't me as executive directory director who said, I've got to find someone to speak on K through 12. No, there's somebody in the community that's immersed and they're going to speak about it just like the other major topics. So it has a great diversity, and great collaborative spirit and energy be as that gets it though, one of the things that's very unique about haystack it's not just the out, put in is open source software. The the process is true open source worldwide community to wrestle down these problems, solve them contribute debate, argue and agree on consensus that's very different than any of the others out there right now.

Phil Zito 55:17
Do you see yourselves maintaining a hybrid model going forward for

John Petze 55:21
the, I think it will always be hybrid, because I don't think you can, you know, we can get back together and have fun and a few beers and have a band again, that'll be great. But you'll never be able to stop the virtual component now, because it serves this other audience. And we've all become more familiar with doing it. And the costs have come down, we looked at adding a virtual component, we weren't going to call it that it was gonna be online component in 2019. Yeah, but the what, what it would have taken to do it cost wise and technology wise, etc. In 2019, which doesn't seem that long ago, it's completely changed. Right? Now, we it's viable. To do that. When we go back, you know, in 2023, to one person, it, I guarantee, it'll be it'll be hybrid, that's a permanent thing. We're doing that in our back on the sky foundry side, we, we before because we're our event is in the fall, we're going to be able to hold an in person event, but it's going to be hybrid, because it will be forever now,

Phil Zito 56:23
it's a much lower barrier to entry.

John Petze 56:26
Yeah, it is. And you want the people there, right? And we're all more comfortable with doing this type of stuff or participating this what this way so.

Phil Zito 56:34
So I want to talk to two audiences here. So someone's stepping into this from the owner operator side, where should they focus?

John Petze 56:44
I think they should focus on understanding the value of interoperable data that they can combine data and why they should have a data strategy. And then taking advantage of guide specifications that will help them specify whether they're going out to a formal tender, or they're just trying to, you know, negotiate with a contractor. This is what we want you to accomplish, we want our data to be tagged with haystack tags. So it's interoperable and focus upon that they don't have to become expert in the details they shouldn't try, they should understand why they want to be able to use their data. And again, I'll go back, Mr. Owner, your data is more valuable than other capital equipment you just put in and let them scratch their head for a moment and challenge it and we can prove it. Right. So data strategy, and in fact, we've had a number of haystack presentations that that is where you need to think first. And when you think about, okay, I care about my data, my data has value, how can I use it? Well, then you fall into? Well, it's got to have semantic modeling, tagging metadata, whatever term you want to use, depending on the audience. That's where they should start is to be thinking about a data strategy and the data they have available. That's one things we haven't seen, what data do you have available? Or available with what level of difficulty, right? And it turns out, there's lots of data that is available easily. And yes, there's lots of data that's much more challenging to get.

Phil Zito 58:16
Okay. So now to our second audience, so contractors, technicians, programmers, they're coming into this, where should they focus?

John Petze 58:26
Yeah, they need to they need to learn the details, right? They need to start with the the entry point, the haystack, the primmer, right? What's it about? What's the concepts we're solving worry about all the gory detail? What are the concepts? How does a basic standardized vocabulary work? How does that then extend to relationships between devices and ontology relationships and those things? It's, it's not something you learn in a single day, right? And I think that's part of the challenge. Why do people think it's hard or it's new? Hey, do I remember I don't know if you're old enough. The transition from pneumatics to DDC, there were a whole bunch of people. I connect pneumatic tubes. What's an analog input? Right? Yeah, it was hard. It wasn't hard. It was something new. You had to learn, which we often you know, quickly say that's hard. Yeah. Yeah. But it isn't hard. It's something new to different set of processes,

Phil Zito 59:26
right and thought processes. And

John Petze 59:27
so it's so they need to learn the basics. And there's lots of materials on that. But we'll admit, we have produced so many materials over the years that it might be hard for people to find an entry point. One of the sections is going to be you know, getting started with haystack, right? Yeah, how do you get started with it? But as a system integrator I the other analogy I'd make let's roll back 10 years ago when the big talk around a lot of the shows was Hey mister contractor system integrator, you need a networking guy. You need to understand networking and IP They didn't before they just connected the manufacturer x networks, whatever it was. And they knew how to do that because of the block. And they didn't know what the data principles were just Oh, yeah, connect this salting, no, you have to understand routers and subnets and VPN. And now any system integrator does it maybe control contractors who haven't bridged that, but system integrators, that there's no argument, you got to have a guy on staff or multiple who know that? Guess what, you got to have a guy on staff who understands data, and data modeling, and data concepts and principles? You have to it's part of our world now. Right?

Phil Zito 1:00:37
Yeah, I would expand on that to say, He is someone who understands data principles and concepts and models, but not a data scientist. We're not expecting someone to go into our and start writing algorithms to consume data sets. But we are expecting someone to be able to analyze an existing site, figure out what the to B data architecture should look like what it is currently, and then figure out a mitigation plan for that. And that requires very little programming.

John Petze 1:01:09
Absolutely, this is a great point, I'm really glad you brought it up, you do not have to be a data scientist to understand haystack and start using it effectively. Right? As I said, you can use haystack with a yellow sheet of paper as a way to uniformly capture important information. So I can give it to Phil, you can do it in Excel, or Yeah, you can do it in JSON and XML and all that you don't have to write the most important thing is understanding the principles of we're going to characterize this equipment and their data with a standard vocabulary, standard organization. So we understand what these things are. And I would say a good test is the human test, can I tag this data, so you could understand what this sensor valve is, what it's connected to, and all of that, it's a great test, you do not have to be a data science. What we're doing, though, is making the information from our systems available for the data science so they can do their stuff, right? They can run their algorithms analysis, etc. Because you know, what one of the greatest barriers is, and, you know, is a great article on, you know, lots of these things around data science for the level I, I can play in weather, and data scientists talking about, yeah, all the algorithms are magic, you know, where we spend most of our time characterizing the data? In other words for tagging it and modeling? Yeah, absolutely provide on fine next grunt work, they hate it. Because if the data arrives to them, with no descriptive information, no metadata, they have to spend all that manual effort, if it arrives, they are tagged, then even if it was tagged with a different standard, you can write parsing routines to interpret it. Ideally, we should all have one, like we have with HTML. But the fact you have any that's consistent and based on a defined notable standard will greatly help us all.

Phil Zito 1:03:03
Yeah, I think we'll get there. I think this should be an every standard for new builds going forward, or every spec for new builds going forward. They should require some semantic model to be adopted, whatever that is. Yeah. Interesting. So I know we're at an hour now but anything else you want to add for the audience prior to us closing out?

John Petze 1:03:25
Yeah, I would say the I would go back to a stack Connect there's no cost to attend the virtual event. The detailed program I mean, there's an overview of the program but the detail program be posted. Think later this week. You don't sign up and come to one or two of them come to the intro ones come to the deep ones, you know, whatever. Start make getting familiar. How do you learn about a subject? Right? Well, you can go to school while we're all working, we have a job what do you do you read? You learn you ponder, you think? Oh, analysis, something came up in my job. They want to talk about tagging? Yeah, I've been studying that boss. Right? I can help with that. Or I want to help with that. Right. This is part of our future just like IP networking became like direct digital control became I guess it was something before pneumatics right. That somebody probably complaint when it changed it right. So there you go. That's, that's what I would say is get involved but don't make it like some huge burden that you have to learn it all at once. Just this is part of the future right?

Phil Zito 1:04:25
Awesome. Oh, perfect. All right. All right, folks. I hope you enjoyed that interview. Once again go to podcast at smart buildings Academy calm for slash 247 once again, that is podcast at smart buildings Academy comm forward slash 247. There you will be able to download the audio if you so choose, you'll be able to get the show notes. You'll be able to listen to the recordings, you'll be able to get the signup link for the haystack Connect conference as well. I encourage you to check all of that out. I will see you next week when we will be continuing our talk on retro commissioning and operations and maintenance. We're going to be focusing heavily on the owner side for the next probably month ish. Alright, hey, thanks a ton. Good seeing y'all have a great West West. Have a great rest of your week. All right. Take care.

Transcribed by https://otter.ai

 

Phil Zito

Written by Phil Zito

Want to be a guest on the Podcast?

 

BE A GUEST