Energy Bytes

Alex Fortin, CTO of Coldbore Technologies shares his unconventional journey to becoming a tech innovator in oil and gas and explains how Coldbore's cutting-edge tech stacks and AI solutions are solving industry-wide problems. The episode dives into the challenges of traditional fracking operations, Coldbore’s contrarian approach to efficiency and sustainability, and what the future holds for technology in the energy sector.

Join the conversation shaping the future of energy.
Collide is the community where oil & gas professionals connect, share insights, and solve real-world problems together. No noise. No fluff. Just the discussions that move our industry forward.
Apply today at collide.io

.

0:00 - Intro Energy Bytes Podcast
2:58 - Alex Fortin Coldbore Background
9:16 - Coldbore Technology Overview
11:54 - Energy Data Wrangling
15:04 - Data Synchronization in Energy
19:22 - Non-Productive Time (NPT) Analysis
20:19 - Edge Computing vs Cloud in Energy
23:13 - Energy Tech Lessons Learned
27:10 - Consuming Coldbore Data
31:13 - Data Management: Shared Rows
32:39 - Introduction to DuckDB
34:16 - Time Series Databases in Energy
40:35 - Cognite Technology Stack
44:50 - Why Cognite Chooses Rust
53:16 - Alex's Favorite Social Media
53:42 - Preferred Cloud Providers
54:04 - Top Vacation Destinations
54:47 - Favorite Programming Languages
55:05 - Best Managed Services
56:10 - Connecting with Alex Fortin

https://twitter.com/collide_ai
https://www.tiktok.com/@collide.io
https://www.instagram.com/collide.io/
https://www.facebook.com/collide.io/
https://www.linkedin.com/company/collideai

What is Energy Bytes?

Welcome to Energy Bytes with John Kalfayan and Bobby Neelon. Your essential guide to the intersection of data and energy. This podcast dives deep into the world of energy, shedding light on how data, AI, and technology are revolutionizing this sector. Each episode equips listeners with insights into the most efficient tools and resources, paving the way for a data-driven future in energy. From technical nuances to broader industry trends, Energy Bytes offers an unparalleled perspective on the evolution of the energy industry. Join us as we decode the algorithms of energy, one byte at a time.

Welcome back everybody. I'm excited about uh today's show. Rad dad in the house. John Calfan. Bobby Nean. What's going on? Um Bobby's sad because Bellich finally retired. Well, you might be happy if he retired. He got Mayo. I mean just he wasn't I mean he wasn't turning the ship around in the next five years, but I mean, you know, we'll see how it goes. And then uh we've also got Alex Foron, the uh CTO over at at Coldboard joining us today. Appreciate you uh jumping on with us, man. Yeah, great to be here, guys. What uh what's the current temperature up there? Because we're currently got that cold front coming across. I'm really bad with Fahrenheit like uh like I gotus 37 Celsius. I know like it's kind of convers around there, right? So I think [snorts] it's probablyus 32 Fahrenheit, something like that. Um so car didn't the Mustang didn't start this morning. Uh so uh got the kids at home. I was like, "No, we're not doing this. You guys staying here." That's [snorts] awesome. Yeah, it really does make you appreciate your uh your natural gas feed out here. Oh, yeah. No, that's uh that's one of those things that I love about the south. Like I wish we had actual winters and falls, but I also like the threat of snow. We'll shut down schools here. If it does snow, we're definitely not having school. Like if it ices, we might be out for a few days. I remember in college we had an entire week off because there was an ice storm. Like and I just went to another college and partied all. It was It was awesome. But yeah, up north people are you just got to go get your [ __ ] done. [laughter] Yeah. Yeah, I know. Be like I'm going to Houston uh next week. I hear it's going to be around uh in the 20s Fahrenheits. Yep. Uh and uh I'm sure it's going to be a bit more chaotic than here in Calgary in the minus uh 30s. Uh surely. Sure. Well, depending on how long you're here, it'll probably be 70 by the time you're gone. So it was 70. Yes. Just I I was hoping to get a little bit of warrant while I was down there. What was it? Did Derek [clears throat] Turk was trying to get me to go speak to the SB thing that they're doing at the end of the month and it's in at Kennanascus, I think is how you say it. I'm not sure. But uh yeah, looks beautiful. Um looks beautiful, but like a SB, you know, we talked about this. They don't pay for you to, you know, go speak at their events. You pay your way. And then b it's like it's Calgary at the uh end of January. [laughter] So, not not my idea of uh fun necessarily. No, fun fun fun travel in snow is never uh never the best, especially when you're flying and stuff, but um no, I mean, I didn't go to PA from like when I was in sales. I would not go to Mar the Marcelus from like November through February because I was like, I'm not going to rent a I might have to rent a car and I might have to figure out if it's four-wheel drive or all-wheel drive, like all that fun stuff. And then just the liability of it. Like, I don't drive in the snow. I know how to generally speaking, but not nearly as good as people who do it every year. Anyway, um well, thanks for thanks for jumping on. We'll we'll get out of that rabbit hole, but um tell us, you know, just tell us a little bit about about your background. You know, we met, I think, initially when I was at Hivesell and stuff, talking to you guys about some of the edge edge stuff, but kind of how did you get get to where you're at today? Were you always kind of into the science and techn or the technology side? Did you like tearing [ __ ] apart and working on cars? I mean, I've broken both cars and computers growing up, so I've always just loved fixing stuff, but [snorts] kind of tell us how you got Yeah. Uh, yeah, a bit of an uh say unconventional track to where I'm at today, honestly. Uh, started, uh, my, you know, going coming out of high school was kind of the kid that didn't know what you wanted to do. Kind of just, uh, floating and having fun. Um, and and went to work for a little bit. And um at the time I worked I worked on the railroad here in in in Canada National Railway and then moved and then also worked for Bell Canada out here which is a big telecom company out east. Uh that's really kind of when I started in technology starting to install fiber optics and you know dabble in everything networking. Uh, and then, uh, was probably a little smart to be, uh, just like a tech out there, uh, full-time. So, they kind of had me do some fancier things. Um, and then started to do some courses on networking and stuff like that. So, we got to do my Cisco, CCNA, like that type of stuff. Um, and then, uh, decided to, uh, go back to university to do my compsai degree after that, uh, when I was in my mid20s. Um, and then uh and then uh got hitched up with a chick and then she uh she wanted to move to Alberta. Uh so I'm originally from Quebec City and so uh so we ended up in Cold Lake uh here. So that's uh that's one of the the that's the big uh can air force base here where you intercept those uh those Russian bombers that comes up the polar circle and uh so I was finishing my degree out there and then that's how uh and then got my first job actually in Edmonton for Coldore back in 2016. Um at the time what we were doing was uh sonic uh like uh acoustic telemetry uh for ball drops. So uh we were like still in incomp completions but on the ball drop side uh right after the 2015 downturn. So it was in an interesting environment. Yeah. Um monitoring for ball drop validation essentially. Yeah. Exactly. Right. So uh so we were you know we're we're developing our own kind of hardware sensors that we're hooking up there. Uh and then we had technicians go out there and what we were doing at the time was trying to uh it was still the early days of stuff like machine learning. So, we're starting to build up a database of, you know, what a does a ball launch sounds like, what do the ball seat sounds like, right? You know, all those those events so that we can confirm on top of pressure um all the the operational during a ball drop. And then uh but then uh you know that was out in Edmonton. So, kind of start kind of tired to do the back and forth between Cold Lake and Edmonton all the time. So, wife found a job in Calgary. I found a job in Calgary and then it split ways with Cold War originally uh to work for a defense contractor here in Calgary. Well, it's an American defense contractor, General Dynamics uh work on uh military telecoms uh which actually is interesting parallels with some of the stuff that we see in the oil field. Yeah. Uh which is like 80s style technologies that uh still run in the 2020s. Right. It's just hard hardened 80s tech, right? Yeah, it's been refurbished and revamped over the years. [clears throat] So, yeah. So, it was working for the British military. They were revamping their stuff, but they're still running kind of their their um high frequency like very high frequency radios from the 80s and trying to send emails over it in the middle of the desert out in Afghanistan. A very interesting uh experience like working there for uh for defense. And then um [clears throat] and then the entire crew at Cobore moved back to uh they all moved to Calgary. We had our sales in Calgary before but all our technicians and and technology groups were in uh Lukes just out of Edmonton and then we uh and then they moved to Calgary um in 2019 and uh and they knew I'd moved to Calgary and they wanted me to come back. So that's when I rejoined Goldberg in 2020 uh about h four four months into the pandemic. minus $49 a barrel [snorts] WTI and uh I thought that stock was a bit low so I said yeah I think it's probably uh it's probably time because I think there's still a lot of growth there to happen and uh and then join when uh the team was still pretty small um and then uh all the way to today where uh we grew the company about uh 10fold I would say roughly since uh since I joined back in 2020 so it's been uh it's been quite the experience. Yeah, y'all have had a ton of success and I as someone who was a frack engineer originally working in the field. Uh y'all are solving a lot of the problems that I was incredibly frustrated with and then moving, you know, moving to the office and Bobby and I, you know, we initially met at Reservoir data systems. So, we were doing surface and downhill gauging gauges with telemetry and that's where I got into the data and the comm side of things. But that was one of the things that we were also kind of I was got moved into product and so one of the new products we were working on was like frag data streaming fragment right and it's like hey these guys are al already doing that yeah that's where we first heard about them it was attech in Colorado the 2019 and I think we're talking to someone with corva and they're like oh there's this company boore that you know and it's like not another well and one thing too is and it's funny too maybe maybe you can shed some light is uh we would always like scout and we would see these Canadian companies and we're like please don't let these Canadian companies come down here cuz like oh yeah cuz we were doing stuff that maybe no one else in our little domain was doing but there were a couple other companies including you up there like please please don't come to America [laughter] please don't compete with us. No, that's uh I mean y'all are solving a huge problem, but can you um kind of jump into just high level, you know, what what Coldboard does, what you know, layer zero I know is a a big thing for you guys. Kind of just talk about that so that it kind of gives the foundation for the rest of the the conversation, so to speak. Sure. Yeah. Um I mean for anyone like to me I always I I try to always draw a parallel back to to what pace on in in the early 2000s right uh on the drilling side and it's really really a good one. Um the the the but you know the reason why completions didn't go through a similar process as fast as it happened on the drilling side is the complexity of the operation on completions on top of the fact that you have uh lots of vendor lots of service companies that are happen to be joined together. No standard with aim for what it is is a standard. [laughter] Yeah. Yeah. So many people just don't realize that right. Like they also don't realize that most oil companies until they start producing the well generate none of their own actual data, right? Like the production data is theirs. I mean it's all their data, but they don't generate any of it until they get to that production phase really. Um and so it's like it's this data nightmare on the back end where you're the operator and you've got wirine data coming in over here. You've got pump down data, frack data, chemical data, sand data, water water transfer data. you've got all of these data sources coming in. There's no standards across any of them. There's no there's not even an agreeance upon like how frequently the data should be pulled. Um, and so it's that's what these guys are are solving. And it's it's a huge benefit because I mean besides drilling frack is the next it's probably I haven't probably on par with each other as far as your costs which are you know that's probably what 95% of any a wells AF is drilling and completions but um we've had all that on the drilling side for so long but on the completion side it has been just a complete cluster. I remember one of the first times Bobby and I were talking to a potential customer about our new frack stuff. He had been working on a frack model, a frack machine learning model to go in and normalize the naming conventions of all their post job data because they, you know, and that was just taking in the post job like thing, not streaming it. that was getting, you know, almost doing their own well well data labs on the back end, you know, 10 years worth of stage post job reports and stage files trying to come in and just normalize the naming conventions so that they could then go build a model to actually do something with that data, right? Like people just don't think about how complex this [ __ ] gets very quickly with all that stuff. [snorts] Yeah. Uh it's a it's an interesting point because uh most of the time like people when they think of coal board they think of uh you know operational we support you know the company man that's out there and and yeah that's a big part of it but but also more and more what we see with uh with clients that are uh you know operators that are starting to get into machine learning and and AI trying to you know leverage that data that they're getting you know a data science project a classic one and that's not just an oil and gas problem it's everywhere problem. It's you get 80% of your uh your time spent on what what they call data wrangling, right? Where it's like how do I put a data set together so that I can now have a clean enough set to be able to do all the fancy things that data science data scientists want to do, right? Um and the the further from the source that you are once you do that operation of wrangling, the harder and the and the most complicated it gets, right? Absolutely. And that's really a big problem we solve for them at Colore because we do this standardization and contextualization of the data um on site in real time. Uh so that you know there's not any guesses that needs to be made post job. Once you know the pad's done like everyone's split off, they're already fracking a new set of wells. And then if you're trying to find out kind of like hey like were we fracking this well or that well at that time or were dual fracking because I see the pressure it's kind of kind of weird there by then like that information is out in the ether and it's really hard to make sense of it. But even like time stamp synchronization, that was one thing that we would pitch, you know, say even at RDS, you know, or especially when we were doing the Frackman thing is because we would have it all coming through the same, you know, DAC or data acquisition device, you know, data logger. Yeah. You know, because we had our our own pressure gauges and then we would stream the frack and you know, they would be on the same time sync and that was at the time where we had like reveal and then you know now you've got um some of the stuff that came out of dev and the steel wellbor stuff where all right well you've got this you know pressure gauge that has its own time stamp and then you would get the frack data and then people would have to merge that and they'd have to shift the data and I mean who knows if they even if it was UTC or was it in the right time zone did they account for daylight savings and then just you know the float of computers is different so like there's no way to even really assure necessarily get that. Whereas, you know, when it comes from a system like yours that it's kind of all coming through the same thing. I'm I'm assuming, you know, the wire line data and the frack van data and then even associated pressure gauges and all that stuff is coming through, you know, the same, you know, computer roughly and you're getting they're all time synchronized. Yeah. I mean, we would have to put pressure gauges on the treatment wells even though we were monitoring the offsets for frack hits just so that we could normalize the frack treatment data or actually it was for the client. The client would do that. So they would use the [clears throat] treatment data from one of the pressure gauges that we had on the treatment well and then they would use that to line up the frack stages and then they would layer on the offset gauges to see if there were pressure hits. But like without that gauge on the treatment well, it was pretty much impossible to figure out how you layer over those stages on top of them [snorts] or someone has to manually go start tagging it. And the way we look at this synchronization is twofold like you're saying. first the obvious one is kind like okay let's synchronize all our computers and damn NTP server so we don't have to worry about that too much and and GPS even like everyone we always have GPS's on site right it's like it's not that hard and that's that's a really good source of time um and and we make sure that we we synchronize all that data for for the customer so that again like when they come on the other side then they can trust that if they want to correlate something that happened on like you said offset wells and and while fracking was going on then they don't have to think about that. Y and two there's also the fact that at Cobar like really believe strongly that you know having the information of the factory uh and being able to correlate all of the operations on pad to what's happening on the factory was also like it's kind of like a it's it's another type of synchronization because really kind of like you you want to synchronize around because essentially like everything that happens on site ends up at the factory right so you can correlate all those activities to the factory Um, and that's that's what we really kind of started making a difference is, you know, back in 2019 there were no like, you know, downing and and that or or they were, you know, not as well known and so like most of the fractures you're seeing out there was this dumb iron and then starting to equip them with with a bunch of pressure transducers and val position sensor allowed us to create this data set that when I talked earlier about the contextualizations like I can Yes, I'm streaming the data to the customer but I can I can also give the data to the customer that already is decorated if you will with like hey you know you valve opened at this time uh so you started fracking there you were idled for about this valve you were you were idled for about 2 minutes where the factory like was ready to get pressure but uh I don't know what the frag guys were doing so for two minutes nothing happened and then we see pressure up and then you know start your slurry then your you know your pad your your flush started at this time. And so when you're trying to analyze something like you even go from that contextualization layer to say okay what am I interested in? Okay, stage 57 will be um you know plug and perf phase. I can I can just ask the system to get the data that's relevant to that and that makes a big difference in the workflow for customers. Yeah, that I I believe y'all were the [clears throat] first ones that I ever saw that had the the valve positioning sensor on it. And I thought that was just such a brilliant [ __ ] idea because it allows you to it allows you to understand to your point all of the context, but also those little very important things like which well, right? Like I can't tell you how many times it would happen where we would put equipment on pressure gauges on a well and the company man would say it was this well name and then we would go back and then the engineer calls us and it's like why is the well name this? it's supposed to be on this well like well this is what the company man told us to do so that's what we did because he's the ber like and then so you you expound that over you know that's the thing with frack data right is it's one it's pretty dense as far as just the volume of it generally speaking not as not like big big data but then you have so many stages right you're talking about thousands and thousands and thousands of stages every single year that you get and so it's like I mean when I was in the field we didn't put the well name in the the the DAS file, right? Like that wasn't in there. It was in the file name or in the folder that the the file was stored in, right? Like and so, well, [ __ ] that doesn't work for a database when you're trying to do big data analysis on the back end because you don't know what well it is. And so, just like those little nuance things are just so important with this type of data, especially on the operation side, because that's where all the money is spent, right? Well, you talk about, you know, I think you were starting to hint some at like NPT where it's like this valve was open and two minutes went by while we're just idling the motors and we weren't actually even doing anything productive. And it sounds insignificant for one, but right compound, you do that per how you know three or four wells on a pad times 30 plus stages crews working, you know, across 365 days and like now you're talking hours, you know, that cost a lot of money. Yeah. No, there's so much those NPTs, too. Like what's great about uh a system like ours is that you can actually tell from the sensors, right? You can have a sensor driven like algorithm that determines who's responsible like in that case of the frack idol like who's responsible from the NPT is it the frack company or is it the the the factory company because you'll you'll see if the valve is open and like or vice versa. Um, and so like normally like you would get if you just look at the data all you see is like I see that nothing like I see I see the well switch took two hours right um so but I don't know why right I don't know why right exactly yeah [snorts] no that's uh [clears throat and cough] like I said it's just it's one of those things unless you've messed with it and tried to do anything with it you really don't understand how hard that problem is is or was to solve um out of the gate But, you know, there's a couple things I want to talk about. I want to kind of get your [snorts] your um kind of strategy as well as just recommendations around, you know, how y'all approach the edge side of things because y'all are unique in that fact. you know, a lot of people, you know, they get data from APIs or whatever, but they're not actually all the way down at the ground level where the data they're not only acquiring the data, but they're generating the actual data and then transforming it and then pushing it up. And so, I mean, I'm like I said earlier, we met because I was working at an edge company. I am pretty bullish longterm on edge in the oil field and the things it's going to be able to do for us, but you know, you guys are actually uh implementing it and doing it on a day-to-day basis. I'm very curious on to as to just you know your your thoughts on it. How you know the things that you guys have have done that were big wins some of the things that you you know you would uh if you were at another service company sitting and watching you talk about it lessons learned things that you wouldn't do and and things like that. Yeah sure. Um yeah it's a very good question. uh like I mean in terms of processing uh we're not I would say that we're not doing so much processing uh that it's it it be because a lot of times like when you hear the stories about the edge is like people start to do so much processing in the cloud it just becomes non-economical data storage. Yeah. Um so for us like we're really um uh doing doing the contextualization a lot of the computation on site uh not necessarily because like it's like a cost saving or or cost uh um optimization path but it's more like that's the that's the the the service that we deliver to the client that's actually on pad right uh so by default we have to do all our computation because like if you look at our edge application versus our cloud application is actually the same thing, right? Um, and the idea was always and especially like, you know, think pre-starlink world. I know like we're all used to having low orbit satellite now, I guess. So, uh, but like just three years ago. Yeah. Um, you know, being working in New Mexico, like how can I make sure that my client has like access to the dashboard uh over the internet at any given time over a cell tower out there? that was not something that we could guarantee to the customer, right? And so that's kind of how we develop our edge solution was the fact that we're isolated with potentially no internet access or inconsistent one or low bandwidth one. Um, and that's much more like what drove our adoption of like edge, the way that we did it, rather than what you hear, you know, the talking hands at Silicon Valley as like what edge is for and why it's important. Yeah. Um, you know what I mean? No, for sure. That's that's the funny thing is like I'm bullish on edge for oil field because of the the connectivity issues and remote piece and the risk. I mean, I can't go blank waiting on it to get to the cloud. You mean you can't have it the cloud be that business critical piece when you're running field operations like they need that visibility and they're paying for that visibility that you're providing them, right? Well, and even when you get, you know, just as something as simple as, hey, the connection went down, right? Being able to have a local device that stores the data and then when the connection pulls back up, has the pub sub [clears throat] relationship that checks and sees, okay, well, this was the last record in the database. Here's all the records I've got. Let's go back fill that so that we can catch up and just keep things going, right? Like just the simple pieces of that are so like relevant to the oil field because that happens all the [ __ ] [laughter] time, right? It's not if things will go wrong, right? It's when and how do you mitigate? Yeah. Um and that's always kind of been our philosophy, right? It's like we need to be able to buffer that data like most of the stuff that we run that feels hardwire like there's so much radios [snorts] and things like that on path and all sorts of interference that you know Wi-Fi even out there like we're it's always kind of a crapshoot, right? So, in terms of our data, like we if if we're going to do some some uh some wireless, we're going to be something out of classical bands and stuff like that to make sure they're always by ourselves. Uh so that that's one of the lessons learned I would say is like you know the your strategy um when it comes to wireless. Wireless is always uh good on paper. Uh but in our experience has been uh has been quite challenging both from a from a performance standpoint and security standpoint. uh when it comes to the big lessons learned for us that we get really nerdy about it um between edge and cloud is is uh this two-way uh you're running an app that uh basically is a duplicate of one another from edge to cloud. Uh we've we've always kind of tried to uh be able to do two-way communication like so like that you have your database synced from um edge to cloud and cloud edge right uh for example like you get um let's say you get a a water like we do a lot of work with water transport company out of Marcelis um like commodities the Marcellis is kind of a pain right if you've ever been in the Appalachins like you know why uh and um and so like they would send us their data you know from uh from an API. So we you know query their API in the cloud and I moved this in our cloud database but then we wanted to drop that down to the edge so that if we had a customer that was relying on our edge device uh application like he could also have all the information he needed. he's actually the one that cares about water transfer, right? Um and uh and and and I would say that that that's a pattern that we've kind of decided to go um and and change because as soon as we wanted to do like uh dual synchronization like two-way synchronization like it opened up Pandora's box of complexity and edge cases um and so that was one of our lessons learned in our in our journey on that is now really trying like we moved all our infrastructure especially now we more good bandwidth on site like to we move all the data to the edge first and then we have to have this oneway pipe to the cloud. Um that simplified our operations our data operations like tremendously. Um and also works with our philosophy of like you know we want the uh the person on site to be the first to know. Um and and so that was one of our lessons learned I would say in the past few years when we're trying to um to pair up and make sure all those applications stayed in sync. Um one way database sync is is much easier than two-way and it's not just twice as hard. It's it's a square rather than a multiplication. [laughter] Yeah. Well, I mean, and let's stick on that, too, because um I think you know, my my team had been talking to you all [clears throat] and then one of our DNC techs left and you know, she was a big advoc advocate for Coldore and just for whatever reason never got legs. Um, but I mean I'm more concerned on the data analytic side of like how do I consume the data from cold boore like are you guys going to generate all this data and it I think it seemed like you guys had two ways for me to get the data. You we could wait for it to get to the cloud and we could access say a restful API but then you also had an MQTT service as well I believe that we could tie into. So if I was going to talk about those options and Sure. Yeah. Um so um I'm going to talk about a lot of lessons learned there. Um so uh yeah so on site uh we do MQTT uh and um we do MQTT websockets right so we have two ways to do streaming on site uh websockets you know it's it's more like the common uh web native approach to streaming uh and MQTT uh because there's so many kind of IoT um type devices that are have that sort of built in uh and then you have the API right and on the cloud we started to do uh some MQTT as well at some point. Uh we're starting to stay away from it because I guess like you need to have sort of a giant MQTT broker where you have to parse everything through. Um and it doesn't it didn't really scale well. Um so um like right now what we're really looking into is like um like you say from the data analyst standpoint where we decided to go instead is more of of uh providing queuing uh through some some CFKA cues, right? Uh so now you have kind of three ways to get your data right you get you can query through an API uh you can sit on a websocket to get in stream in real time but you have also the queueing which is kind of annotate about those things because the the problem is always like well an API if I give access and everyone wants to do second per second data on my API they're crush my AWS right um which obviously is a problem uh streaming well from the client side it's like well if my client dies for an hour and I don't realize it then I have to do API queries and then like all of a sudden your client becomes super complex uh because for failure uh you know error recovery uh there's so many edge cases that pile on to make sure that you have uh you have all your data. So what we what we went to is two two two other options that are kind of a high a number of that like firstly CF Q. So that way like we drop the data like we standardize it. So like basically you have you end up with a bunch of cues that are your standard channels that you want from your frack, right? Uh so for each job, each standard channel and again like whatever service companies out there, whatever operator you are, these things are always named the same. So from the operator standpoint, like you don't really care if it's Liberty or Albertton that's racking, like it's always going to look the same for you, right? So that's another big piece of what we do is we create this abstraction. But then with the queueing, yeah, [laughter] that's that's just an industry problem, right? Like I'm going to overall. Yeah. Like every everybody should try and standardize naming conventions, at least on the services side. I like I think it's crazy to me that as an industry, we don't just have a generic naming convention that most of us agree upon. We don't even have that. Like we don't even care about structure or any of like the file types or any of that [ __ ] just like a naming convention that is standardized across frack. We don't have that. Like companies do, but as an industry, we don't. And it's that's crazy to me. [snorts] Yeah. And so, yeah, going back to the the queueing. So, you you have that those standard channel then and then we can queue up to like, you know, six hours worth of data for each channel. And so, then the client can uh kind of go there and uh consume that as as they want. Um, and that's that's that's been a that's something that we're rolling out right now to some customer that's been very good uh in simplifying the workflows. And then you also have uh we do uh we work with snowflake so or whoever uh other type of snowflake like uh thing to be able to you know do those shared rows uh like snowflake share now where you can just like share it across like three you're talking about. Yeah. No, that's He loves a parket file. Well, not even that. I mean, we don't need that. I mean, literally like if they have Snowflake, we have Snowflake. You give me your account ID, whatever, and bam, their data shows up as a database in my warehouse without ever having to do any Yeah, you can literally join to our database essentially, right? [clears throat] Because you see them just like there were rows in yours. Um, yeah, once it's enabled, like it shows I mean, we actually have it with Liberty and like it just shows up across the wire and now it's part of your database. we should just get rid of APIs and that's all it that's I mean that's it's moving there. I mean it'll never go away but it's moving there like I mean what [clears throat] Salesforce and AWS and Salesforce and Snowflake have it where it just it automatically syncs. It's just there because then like you know you're you know we all spent our time like you know we're all in the same either you data center here and then we keep going over the internet. Yeah. [laughter] Right. And then like the Snowflake guys are just like, "Hey, what about you know we just stay in the data center over that big fiber optic as big as this, right? And just do a share of our SQL query our SQL table." So that's that's pretty that's that's super fast. Yeah. Now that you know storage is decoupled from the compute now it's just you're just sharing their file structure across you and then it just works. Yeah. Oh, that's God. Yeah. It's a pricier model. Um so we'll see. I'm excited to see a bit of competition in that space. Uh Snowflake is obviously the big one. Um the other one is data bricks which is also a big one. We've been kind of interacting with uh both those platforms. Um but there's a lot of um and we've [clears throat] worked with customers that have started to look into because there's some open source project that are starting to crop up on kind of the same patterns uh that will make that a little more economical I think for everybody. So uh that's going the right direction and house. Yeah, we see uh how do they call it? Um name name escapes me. There's there's one of those was open source project that's very popular like um yeah something share. Uh okay. Yeah, I'd have to look it up. Yeah, I mean I was saying duct DB or click house. I mean just as far as those are right those columner, you know, um engines that are that compete, you know, on the same scale of a, you know, snowflake or whatever. But again, they're um but I mean even that I mean I don't know. I I guess I'm I'm a duck DB bro who hasn't even really used it that much. But like um you know curious I'm always trying to see if people have even messed with it because I mean like I think that's one of the things that's so cool about that is it's SQL light for analytics. Like you can run duct DB in the browser, you can run it on the edge, you can run it in your data center, you can run it, you know, just about anywhere. Um so it's pretty cool, but I don't know if y'all are using anything like that at all or messing with it. Not really. Not right now. Like one of the main challenge on the database side for us has always been the time series stuff. Yeah. Right. Uh the way that we handle time series data. Um because uh it can get get really crazy as long as soon as you have like engineers that are starting to pound at your your web application and look at charts over like oh I want to see the frack chart over like the last 15 days. Yeah. Right. That's like you know [clears throat] 10 million data points that I want to throw at the browser. Right. So yeah, Bobby and [laughter] I Bobby and I have had many I mean because we were doing like mostly defit or frack analysis, right? So it's all one second data, but it's normally during, you know, a defrack or whatever, but even during the frack, right? Like they're looking for frack hits during the entire frack job. So they want to see 30 stages with 30 potential frack hits across it over a twoe period. And we've had many conversations, Bobby and I, about like with potential customers or engineers of like you physically can't fit all of those data points in the pixels of this screen. So we cannot show you every single data point. I know you want to see every single that you think you want to but we had some like smart little algorithms and stuff that were kind of windowed as you zoomed in and out that would kind of expand or contrast and I think that's what most people end up. But even that to hit you I mean they'll still have to interact with the database and you need to have a database that's able to handle that. So I mean as much as you're willing to say I mean I mean I don't know I mean um you I'm curious if you know what technology you've landed on or do you kind of just homegrow the way you're ingesting stuff into a relational database because I mean I know you know Postgress has time scale or there's influx and there's all these different time you know kind of time series databases. So curious what the journey has like there. Yeah we use time scale. Uh I always been uh not a big I'm I mean I've worked with SQL for so long like I I've kind of like we ended up there because it's something you're like super comfortable. It always does. Yeah. Yeah. It does always works right. Uh so times and uh it's been very successful like you're saying like it's kind of like you have your built-in uh ways to create samples of certain buckets of time and then you you can adjust the query from the web application and stuff like that and then really able to get that level of performance that you know from a UIUX standpoint you want to be able to provide the customer. Uh I would say there's been a bit of learning in terms of the you know the way we're sampling things. Uh, I mean I'm I'm more from like the finance side when it comes to like I mean I'm kind of a finance bro. So like I'm I the first time I looked at a frack chart I was like how the hell does people like how do you look at this like [snorts] you know uh and and so there's been a bit of learning on at least for me and and a lot of the other engineers that I'm working with that are not from the industry as to like what a frack engineer expects to see. Right. Right. And how we sample that data so that you know it looks right. um when when we do some some downsampling of data. Um because that was some one of the issues that we had earlier on was that the way we were down sampling was not like they were like yeah if I zoom in it makes sense but if I zoom out I'm looking at this chart and it looks wrong. We felt those we know exactly what you're talking about. [laughter] It was funny too. I mean like I don't know like even just listen you guys arrived on time scale and John and I went through things. I mean I look at there were a lot of architectural decisions that I wanted to be able to do but it was but and what is what I would tell our CEOs like I want to do these things like what you realize and you want us to compete with well data labs or cold boore who cor like it's like it's me versus like 20 Ukrainians like and [laughter] yeah but I mean like but I hear some of these like man we were on the right track you know like we just didn't have necessarily the resources whether it was time people I guess it's you know or even experience too or the experience I wasn't point then you know I [laughter] gone through a coding boot camp right no that's yeah that was a that was some learning for me too like even the way that those time series database like influx and and time scales as time scale have evolved over the past like just three to five years is is massive it's just massive and I I'm like like I think we're I'm starting to understand like yeah that's why only the big ones were kind of doing those type of like high volume of data applications because before that we don't have any great tools to be able like as a you know a small shop to be able to to do that by ourselves. Uh which surprised me that even like that you know in in the late 2010s early 2020s we're still kind of at that phase in terms of the the stack. Um but it is true. uh it's only I would say in the last 3 years that we have we have really seen some um some I mean they're open source and enterprise level so there is a premium type of uh of um of of licensing but that's really kind of like in the last 5 years we see things that we can use at scale uh at enterprise level and and not have to deploy like a team of 25 engineers to tweak and manage it inside your organization especially like time scale I think I mean time scale was the right call them. But like at the same time it was before they there was really any managed time scale services. So then like it was you know our you know the contractors we're using were standing up Postgress you know with time scale on a VM on AWS which a is expensive and b like they didn't even really know how to set it up properly and optimize you know so you know which is one of those like at a when you don't have the time or people or resources to really dive into it you know time scale uh they've just released their own managed service like they they they were outsourcing their managed service to another company that was hosting their open source thing like yeah it was Ivan out of Finland right they were doing that for them now they've just moved their stuff out in north back to North America and um and but but that gives you an an idea of like how still pretty young you know that that all is interestingly enough I think you could spin up a time scale Postgress on Azure before you could do it on AWS like you know they've got that as the managed service but I want to know just as much as you're willing to kind of talk about or can talk about kind of what what y'all's stack looks like specifically around are you using like containers Kubernetes and then I also heard you mention CFKA like I think CFKA is a really interesting tool that some people know about in the industry but a lot still don't and then you've also got you know Confluence you've got the manage version of it and stuff like that so I'm very curious about kind of you know what y'all's architecture looks like and and what you know things y'all have been able to to do to help like just containers, right? Like when I was selling [laughter] edge, uh the amount of people that are still running VMs for so many things is just and it's not just oil and gas, right? It's like pretty much every major industry has a lot of very big VM players that have always been built on VMs, so they are still built on VMs. Um and so like just things like that, right? Like that I'm very curious to kind of to hear like [laughter] and again you can't share or don't want to or whatever, that's fine, too. [snorts] No, I don't I don't think, you know, there's no big secrets there. I mean, we're we've always been running a Docker stack for both uh both on the edge and cloud. Uh like we have this micros service type. Yeah, we have that microser architecture stayed away from Kubernetes. Um I mean like I was working with um one of our contractors. He's a guy from uh uh he's a guy that was working at YouTube before, a guy from the valley and he worked with me uh when I started. He was one of our contractors and he was part of he was kind of adjacent to the team was working Kubernetes and he was like dude like the amount of people that actually need to use Kubernetes is really [ __ ] small, you know. [laughter] No, I I wholeheartedly agree. Like that was in my opinion one of the main kind of negatives about I thought it was a positive going in because I was didn't know better but about the the hive cell setup was that you know the Kubernetes offers that resiliency and high availability that at the edge we think we would ultimately want but then of course no one actually wants to pay for it until they have that issue where oh well we're in the middle of a frack stage and all of our equipment [ __ ] the bed and now we don't have any data and it's like well is it really worth that like mitigation or do we just have a backup sitting out there that we can plug in and keep going, right? Especially just the the expertise around is still so small like Yeah. And and that's the thing about engineers like when we work with engineers um software at least that I can talk to that because that's my world but like it's like you know they they're attracted to complexity like moth to a flame right uh and so like everyone wants to do hub enemies because like that's what the cool kids do type deal like we've we've been running our docker compose stack for forever and we've like we know everything that used to know about it and we never had any issue like it's not even on the radar of the things that we're worried about honestly. Uh and so so we've been running that uh we we do some hosted service like we we leverage some hosted service on uh on the cloud. We're um I mean me and Cooper Cooper Grat is our uh uh director of technology. He's uh like we work together for a GD together and all that and we're kind of Linux geek, right? So we're a bit more like old school when it comes to like our run our stack in the cloud because I'm like because we know how it works, right? we we use less, let's say, managed service than maybe we should, but it's more of like like we know how they work. We know what we need. Uh, and we're kind of Well, that's the thing, right? It's like when when there's too much magic AWS [ __ ] uh, I'm like, like if it doesn't work, we're like, I don't know where to start. Like, I don't know what AWS wrapped this in, [laughter] right? No, I mean if I run an EC2 instance and um and I'm, you know, I have my SSH tunnel to to my to my logs, I can probably do a lot of things. Uh so that's a bit more of our philosophy because of our background. Like I understand why it's there, but um I we're more kind of like the the technical guys that want a little more level of control. Uh so so we do some manage stuff when it makes sense, but uh not kind of like a as a default. Um what we've uh been very different about is that uh we've built a lot of our software on new technology. Uh so for example everything that we run on site um when it comes to embedded is all uh pretty much all rust um because uh well again it's back to the story of Cooper and I working for the defense industry in C++ and you're like hey we're not doing this C++ thing again. [laughter] No uh we get the benefits of C++ but with more type safety and everything right on on Rust. Exactly. the type and mainly for the memory safety. Well, not type safety memory. Sorry. Yeah. Yeah. The the memory safety is such a blessing like and everyone that says like well it compiles more slowly like I I you really need to revisit your assumption about what's important. uh like if it takes five times the amount of uh compile time up front and I'm going to save like 16 millions hours of debugging a 12 red process that crashes every two months like yeah like I'm going to pay that price up front anytime right so that's that's one decision that we made and it's been great because first it it it simplified some of these things um and also attracted talent right uh like we had guys that uh And we had people that wanted to work on Rust in production. A lot of people were doing this and that was back in when we started in 2020. Uh and now like looking back I think that was a great decision because you look at 2024 now and you know Rust is part of the Linux kernel. Um we see Microsoft starting to migrate some of their uh CC++ to Rust as well in the uh Windows operating system. So it's great to see because like sounds like this is not going away anytime it's the language you know of choice for all like the um blockchain. Yeah. A lot of the crypto that kind of stuff is in rest and everything above like more on the cloud side we've been having a lot of emphasis on using elixir and OTP. So for those that are not aware like that's all based on Erlang. Uh I mean airline was built by Ericson for telecom, right? And you can draw a lot of parallel to what Cobalt does at the end of the day. Like we're moving a lot of data. Yeah. Uh in all sorts of ways really fast. And um and functional programming is is is a pattern that you know we really kind of embraced uh when it comes to how we do our development, how we think about our our pipeline, our our software, what goes where uh in our architecture. And so that been kind of like our ways to think about it software like the the the the language and the framework really fits well with that. Uh and everything on the kind of the website has been uh a react classic react JavaScript type stuff. Are y'all using CFKA or are you using like Confluent or Yeah, we're using CFKA for for our queuing uh but outbound. So um we've developed our own uh pipes uh with elixir and and the OTP framework for everything that is uh moving data from edge to cloud and stuff like that. CFKA is more like a way to because we don't like there's a lot of CFKA client, right? So it's easy for a customer to be able just a CFKA client and read from a Q rather than you know uh have to write a client that uses our our special sauce. So it's not it's not critical to your application. It's more of a way to save serve data to others. Yeah. It's a distribution. Exactly. Like a neat way to distribute the data to in a queueing in a queue uh to to uh to the end user. Yeah. No, that makes I mean that's that's what they really push CFKA with, right? Is data in motion. And again, there's another one I was I was in I was on CFKA testing early [laughter] like I was using Confluent to connect to our MySQL and it was going to go somewhere else and then, you know, again bringing enough contractors and they overrode that. But um no, I mean [laughter] CF's That was in like what seven 2017 because it was caf cafe connected come out. So I was like oh I can just write a little uh configuration file connected my SQL. I just had to turn on the bin log and just all the data was flowing out of our you know time series stuff and I could send it anywhere. And then um but yeah I was just a a little wet behind the ears and didn't you know didn't really know what I was doing yet. So but there's a lot of things I was like oh that's probably the right technology. Yeah. uh I mean those things like it's it's I mean it all these problems have already been solved right so it's like let's uh yeah it's more about the technologies and creat for our stuff like we've been leveraging like when I say elixir like it's just that it's more of the framework also that we adopted it's not like those uh that data movement and pipelines and and cues like all the all those things already built in the language and so like you just go with that and you don't take the risk of having a guy reinvent the queue and go through the same hoops that the guy in 1952 had to go through. Right. Again, even you brought up M Well, we brought up MQTT earlier and seemingly it didn't necessarily work exactly how you guys wanted it to work, but I think it was Colin was telling me like I think MQTT was actually born out of like Philip 66 or or Ko Phillips or one of those companies. Apparently, there were a couple guys there who developed the MQTT protocol, which I think a lot of people don't realize it was born out of oil and gas. I didn't know that actually. Yeah, I didn't. It was like probably a month or two ago. He told me so. No, I would have easily called that that was a valley invention. But again, but the valley doesn't have to solve that. You know, they don't messing with the edges nearly. I mean, I say I guess phones and stuff potentially, but like yeah, it's born motherhood's the or necessity necessities of the motherhood invention. So, yeah. No, that was what that's true. the valley like when you when you hear the guy of the valley talk about edge talk about modern computing and stuff like that it's really hard to relate to what we do here [laughter] a lot of times right uh like we're in a B2B also right so a lot of the stuff is built around more the B2C type even if you think about product development right uh most of our you know you you hire a bunch of of product manager and they kind of have this B2C mindset of like you know we're going to we're going to try this on the customer we're going to AB test this It's like you're not doing AB tests between KO Phillips and Exxon Mobile. I'm sorry. Right. Yeah. [laughter] So, so like that's that's a very different beast like just from that angle. And then and also like oil and gas I think uh in terms of like adoption of software adoption of technology like I think you know it's a laggered as an industry over like in general uh and so it's always kind of interesting to have to interact and bring new people in our company and to have to interact with a client because like you know blows their minds. It's not at all what you've what you've seen before. Okay. Yeah. They're a publicly traded billion dollar company and they're still using my SQL. Yeah. Whatever. Yeah. [laughter] Oh, that's Yeah. Well, or like you know, Mudbus, like make it that, right? It's like Yeah. [laughter] Modbus, man. I love pulling things. [gasps] Bobby spent so much time with with all the the intercol mod serial. I mean, all the good old Yep. No. Um Well, [ __ ] man. That was that was a pretty quick hour. We're gonna we're gonna need you back on at some point. We're we're gonna try and do in the future some uh round table. Yeah, round table topic specific episode like AI if we get him in there with uh Feder. Yeah, that' be it' be a cool one. Uh it' be a cool one to do is uh if we we talk a bit like if you want to do a bit of a session about like next five years like what this looks like what it needs to look like where what success looks like for uh for like for what we how we see the industry going and and where we fit and and uh that that's something that I've actually done like I was on a panel on SP uh out in the Marcellis just in in the fall and there was a lot of great discussion there uh around that and so I think having Yeah. And like you're saying, it's sort of round table type stuff. I think it'd be great. Yeah. No, [snorts] man. Well, we uh at the end of of each episode, we do a little speed round where we kind of just pepper you with random short questions and just give your quick and short answer and then we'll we'll wrap things up. All right. [snorts] So, um who's your favorite, uh social media follow?

Um wow, that's a good one. It don't have to be like an influencer. can be like a tech guy or just a interesting person. I would say probably I'm really enjoying following Elon around on Twitter. That's two for two today. Oh, that's always I mean he's just fun to watch because all the stuff that's going on. But um what is your favorite cloud preferred if you had to pick one? If I had to pick one, um Azure. Okay, interesting. because it could sound like you all do a good bit on AWS, too. But we do both. Yeah, we do both. Gota, that's how most people should be doing it. Yeah. [laughter] Yeah. You can't lock yourself in. Um, what's your favorite place to go on vacation?

Florida. Okay. I mean, honestly, like, you know, since I went there a handful of times, every time I go to Florida, I have a good time. Like, people people [ __ ] all over Florida, but Florida's blast. Yeah, [laughter] it's it's so nice to go sit on a beach and it's a change from here and especially today. Yeah, I was going to say [laughter] I can't imagine having like it I literally can't comprehend how cold it is there because I've never been that cold in my life. Yeah, I've gotten zero Fahrenheit surface temperatures with 20 mph winds. But that's not it's still not even comprehensible. Um what is your favorite language or um coding languages? either language or like library that you've kind of come across that might be interesting or relevant. Yeah, it' be elixir llang for me. Nice. One more. Um yeah, what's your what's your favorite managed service? That's a good one. What do you mean by managed service? Oh, like I mean like a third party like a third, you know, I mean like maybe Ozero or those kind of thing or definitely not Ozero. Well, you know, John doesn't like through it. But I mean, not just like or you know, uh I mean for me like it's uh GitHub, man. Like I mean that's like so much of our workflows and it makes so much sense from a developer's mind. Uh right like I do everything now on GitHub. Like I do accounting, I do like everything in that workflow. So that'd be that's kind of the one of the things that's open in my face like 247 almost. No, for sure. I mean, I I started diving into GitHub actions over the winter break and pretty awesome what you can do with that. Exactly. Yeah. And yeah, like now my brain just keeps going back to like, oh, I'll just make it a GitHub workflow. Yeah. [laughter] Done. It's smart. I mean, you got to got to do it where it's most accessible and easiest to do. Um, well, where can where can people find you? How do they get in touch with you or Cobbore? Uh, Ezus uh is uh find me is on LinkedIn. That's probably like where uh where I do most interaction uh on social media. I have a I have a X account too, but uh kind of sparsely there. Uh and yeah, coboard technology.com for uh to to get a sense of everything that we do. We'll be at HFTC uh right around what is it? Three weeks from now. Yeah, it's normally end of January, early February, end of January. Um so we'll be there um so to see uh a lot of people from the industry uh there in Houston. Yeah, that's always a great show. Everybody, if you uh you like this episode and like this stuff, check out this episode. Um see what uh what else we've got. Alex, thank you so much for jumping on with us. We really appreciate it. Stay warm. That's fine, guys. Awesome.