Optimizing Smart Meter IIoT Data in Databricks for At-Scale Interactive Electrical Load Analytics

Summary

• IIoT at scale: Octave application analyzes smart meter data from 4.5M meters across Hydro-Québec using Databricks, processing over 1 trillion data points.
• Interactive analytics: The Plotly Dash application on Databricks supports 1,000+ technicians and engineers performing real-time electrical load and voltage analysis.
• Enterprise optimization: Databricks addresses massive scale, governance, and security requirements for interactive meter data analytics.

Optimizing Smart Meter IIoT Data in Databricks for At-Scale Interactive Electrical Load Analytics

Talk by David Gibbon, Senior Director - Strategic Partnerships, Plotly
Octave is a Plotly Dash application used daily by about 1,000 Hydro-Québec technicians and engineers to analyze smart meter load and voltage data from 4.5M meters across the province. As adoption grew, Octave’s back end was migrated to Databricks to address increasingly massive scale (over 1T data points), governance and security requirements. This talk will summarize how Databricks was optimized to support performant at-scale interactive Dash application experiences while in parallel managing complex back-end ETL processes. The talk will outline optimizations targeted to further optimize query latency and user concurrency, along with plans to increase data update frequency. Non-technology related success factors to be reviewed will include the value of: subject matter expertise, operational autonomy, code quality for long-term maintainability and proactive vendor technical support.
Intelligent analytics for everyone: https://www.databricks.com/product/business-intelligence
BI meets AI: https://www.databricks.com/resources/ebook/business-intelligence-meets-ai
Explore the only unified platform for agent systems: https://www.databricks.com/product/artificial-intelligence
See all the product announcements from Data + AI Summit: https://www.databricks.com/events/dataaisummit-2025-announcements

FAQs

What is Octave and what does it do?

Octave is a Plotly Dash application used by 1,000+ Hydro-Québec technicians and engineers to analyze smart meter load and voltage data from 4.5 million meters across the province.

How much data does the system process?

The Octave application on Databricks processes over 1 trillion data points from smart meters, enabling real-time interactive analysis at massive scale.

How do I share internal apps securely with the right people?

This video showcases Octave, a Plotly Dash application used by 1,000+ Hydro-Québec technicians to analyze smart meter data on the Databricks Data and AI platform. The solution demonstrates secure access control for analytical applications at scale, ensuring that internal tools are accessible only to authorized personnel while maintaining governance and audit trails for regulated industries.

How does human-in-the-loop analytics work?

This video emphasizes human-in-the-loop interactive analytics, where domain experts review data to identify patterns and anomalies. The Octave application enables technicians and engineers to interactively explore 1 trillion data points from smart meters to find optimization opportunities. This human expertise is essential in domains like electrical grid management where context and domain knowledge are required to interpret data correctly.

How can I track which users are accessing specific data?

This video demonstrates that when operating analytical tools at scale on Databricks, you need comprehensive auditing to track data access and usage. The Octave application's backend on Databricks enables detailed tracking of which technicians accessed which data, supporting both security requirements and operational autonomy while maintaining compliance with regulatory requirements in heavily regulated industries.

Full transcript

All right, it's 5:20. I'm going to be prompt. It's the first time in my life I've ever been prompt. And I'm going to start this presentation. So, first of all, thank you for joining, right? It's 5:20 on a Wednesday. You've endured two days of this stuff. So, the fact that you're even here like is a bonus. So, like I probably can count you and you know, give you all each a beer. Um, couple disclaimers, right? So this is a talk about well in terms of the title interactive keyword that's why it's in bold load and voltage analytics which speaks to the fact that we're talking about electrical uh you know energy right it has a data bricks twist the d the data that's related to this is coming from smart meters right that uh you know installed in various you know homes and industrial industrial spots and they're looking to optimize it.
Why would you want to optimize it? probably because there's a lot of it, right? So, this is a bit of an at scale. This presentation is not the following, right? There's no AI. So, no AI, right? It's kind of old school, you know, but there's a lot of data, right? So, hopefully you learned a bit about that. Second thing, I'm a non-electrical engineer. So, like the last time I really like dove into this was like second year physics and that's it, right?
So, I haven't touched this stuff for a long time. Second, I'm not a data engineer, right? So there's some slides on here that speak to like data brickscentric you know terminology. Some of you maybe all of you are going to know this better than me. Third I'm not a software engineer right so I work for a company called plotly. Anybody heard of you us all right more than a few hands. So it's great. So we're known for a lot of different things.
Big picture data visualization. What does that mean? Well, oftentimes it means there's a human in the loop who wants to look at stuff, right? You know, and so it's not strictly backend and they want to like look at stuff and in this context they want to do analytics. So that's the preamble. I want to thank a few people who aren't here right for uh for affording me the opportunity. One is actually a number of people at data bricks right who um were involved in this in different way shape or form.
Uh this is I think it's a really good story that's been needing to be told for a while. Uh this was actually originally told in in Montreal in late November and I think it was actually warmer there then it is here today. So you know do that what you may. I want to thank Hydro Quebec. Uh they're not here for certain reasons. I won't go into them but uh I'm presenting on their behalf uh today. And so it'll be less technical.
This will be speaking more about the why and the when you know of this of how this circumstance arose and the problem set and the resolution, you know, versus so much the uh of the how. So feel free to interject, you know, hopefully we'll stay on track here. There's nobody following us. Um so hopefully you'll you'll gain a bit out of it. So with that, I'm going to context switch very briefly and I'm going to go to this.
So this is human in the loop interactive analytics, right? And what you're seeing on the screen is a chart. It's a chart. It's a line chart. It's got two lines, right? They're kind of moving around on the screen, right? I want you to tell me what this means. Any guesses? Ah 24 hours. Okay. Keep that in mind. But that's not the answer, right? But you are on the right track, right? You know, and the reason you're on the right track is we're doing industrial IoT, right?
So, if I look at the next clue on this thing, right here. Oh, my Wi-Fi is off. All right. So, hold on. phone. Interactive means you actually have to have network connectivity. All right, here we go. Okay, so time, you got time axis. So, Good. But that was an easy one, right? The next thing is actually here's the second clue. The y ais is water consumption in millions of liters, right? Any clues? Any any guesses now as to what this is?
Yeah, you're on you're on the track, right? Let's give you another clue. Right. So, this goes back to your first point. This is during the course not of weeks or months but of hours, right? Millions of liters and then some like spikes and low downs. Any guesses? Irrigation. No. Home usage maybe. Right. Next clue. Any city in Canada and two dates? February 28th, 2018. February 27th, 2010. No. Oh. All right. Can't be any Canadians in this crowd, right?
People go to the bathroom not during commercial breaks but during right in between periods you know when this event right golden goal Sydney Crosby 2010 right now why did I use that you know well that's water not electricity Right? It's industrial IoT. You know, it's a scenario where unless you actually really know like, you know, the context and the details, AI is not going to answer it for you. Actually, you know, what I did, you know, was I uh actually went to uh got this off the worldwide web as some of you might still call it, right?
And it's based on this graphic, which will come up here in a second. from EPCORE, Edmonton Water Utility, right? And they mapped this thing out, right? And then what I did is I fed it into CPT and I took away all the context and I added it. It could it did not guess it. It guessed partway through that there was some desalination plant in Perth in 2010 on that date that had some issues, right? But until it got the last clue, you know, it actually told me that there was basically yeah, this can't be due to like regular usage.
It had to be due to some like extraordinary, you know, you know, skate system, you know, in event. So why do I bring that up? Right? You get these circumstances where AI cannot predict it. You got to be a domain expert. you got to have the context, you know, and then from there you can apply that, you know, tribal knowledge, if you will, you know, to the thing at hand. So, and by the way, there's a hockey game going on for anybody who's kind of interested, uh, from Florida, from Edmonton.
So, you might want to tell me how it's proceeding. All right. So, now I'm going to go back to my presentation and continue it. Am I there? I think I am. Yes, you're seeing that. Okay. So, now back to the regular schedule programming. All right. So, this presentation, the feature presentation is actually because it's more about the why than the how. It's about this gentleman. And he's probably going to give me grief for putting his picture on the screen, right?
But this gentleman's name is. He's an electrical engineer. He's in he's based out of Montreal. He works for Hydro Quebec. And he actually is like this renaissance guy, right? He can he can he's an electrical engineer who codes, right? And if you look at his LinkedIn profile and you translate it into English, that's basically what comes up on the right hand side of the screen, right? He's your go-to guy. You got a special project, you want to, you know, get something done, can't quite do it with normal mode operation, he's your guy, right?
And who are his this protagonist's assets? Well, they're people in the field, right? you know, these operatives like their field crew, their planning teams, their maintenance engineers, you know, they're getting stuff done, right? You know, in this in this utility, right? And who do they work for? Wow, they're all like these specialists for Hydro Quebec, right? Hydro Quebec, big utility, world's third largest hydro utility, right?
For hydro power generation, transmission, and distribution, right? They actually what's shown on the screen is the world's largest multis something butress uh dam world's largest dam of some very specific structure. Also Quebec is a big province, right? Big province. Spatially it's about the size a little bit smaller than Alaska. So Alaska is a little bit bigger, right? But Quebec is monstrous. And like and one you know one thing to note is that when you're generating hydro hydro is often if you want to generate a lot of hydro right you got to have a lot of water and a lot a lot of those places with a lot of water are a long way away from where the people are so the generation is a long way away from where the people are thus you need to transmit a long distance very long distances right and so hydro Quebec doesn't have the longest you know transmission lines in the world but pretty pretty long.
And they were innovators back in the 60s of creating, you know, uh ultra high voltage AC transmission lines. I learned so much in my job. This is awesome, right? You know, 735 KV AC transmission like they were the pioneers of this, right? So the Brazilians, you know, and the Chinese, you know, are benefiting from, you know, some Kebeekqua, you know, innovation, right? Now the numbers, right? You know, utilities are like uh capital intensive, right?
Lots of assets on the balance sheet, right? Lots of capital investment, right? Lots of revenue, pretty good income, right? Lots of employees. Those 22,000 employees represent the coal, you know, from generation, right? Transmission and distribution. And they, you know, big numbers in terms of terowatt hours of actual, you know, power sales of which 10% roughly is actually consumed outside the province of Quebec. Some a little bit to the west, you know, in Ontario, right?
Another big province in Canada. Some a little bit south, you know, to our neighbors in the United States of America, right? And these businesses are split into like three big pieces, right? There's the generation part of the equation, right? In this case, it's predominantly hydro. Although if you look in the fine print, they actually, you know, essentially aggregate uh, you know, renewables or like you know, people like selling their power to the grid, right?
So they're de facto. The transmission lines we kind of talked about. This talk is about neither of those two. This talk is about the distribution, the last miles, the last kilometers if you will, right? Of which there's 230,000 kilometers. Do the math. is about 140,000 miles I think right somewhere around there right for 4. 5 million customers population of Quebec is about nine million so like one of every two you know two two people per household so to speak and you know max consumption yeah sorry Mr.
Jacksonville, it's not, you know, in August, you know, it's in February, right? Winter is the, you know, peak season, right? You know, that's when you want to avoid stuff, you know, because like people will, you know, it's cold, right? And here's the theme of this talk, like how does this problem about scale come to actually pass? Well, this talk initially, right, is a is about a great British word. Any Brits in the crowd?
No Brits, right? Bespoke. Bespoke is this great British term for custom, right? You want a very specific, tailored, unique solution for your specific circumstances. One word to describe it is bespoke. And in this particular context, the scene is, imagine the following. You're somewhere in Quebec. It's winter. It's freezing rain, right? Electrical lines are down, right? People's homes are freezing. Why? Because in Quebec, you know, most of the people's homes, 92% of the homes are actually heated with electricity.
You know, the rest of Canada and a lot of the US it's natural gas or whatever else, right? So, if they if the power goes out, yeah, you don't have lights, but you don't freeze, right? You know, so it's a kind of a got a compounding effect. You're who are you in that context? You're a line worker and you're looking to restore power. And as you do so, you know, what you're often offered up by your, you know, management team or whatnot is some app that tells you when you've done your job whether to flip the switch and turn the power back on.
And most times it's a yes or a no. Basically a thumbs up, thumbs down. But here's the butt. You're a grizzled vet, right? You got fat fingers. You got a hard hat. You've been doing this thing for 20 years, right? and you're not so sure in this particular case, you know, if that yes or no is right, you know, and the thing is you don't want to just trust it. You know, you're more like. Trust but verify, right? And so what can you do?
You know, so that especially because you don't want to have to go somewhere else and come back, right? You got too much things on your work order already and you don't want to have people left in the not just the dark, but freezing in the interim. Not a good situation. So here's the what if what if there was a more sophisticated tool that you that person tailored for you could take advantage of all your like tribal knowledge right maybe not to give an exact better answer but to give you a better informed answer you know in that particular circumstance so that's the mission right interactive smart meter analytics and specifically looking at load and voltages because it's actually not specifically the the smart meter that you're really worried about.
It's actually the status of the network in that vicinity right and that's a you know and you can inform yourself you know of the load levels of the variance of the whatever you know based on those as inputs right and you know but also that network is a living kind of organism right you got people who like have just recently plugged into the grid you've got transformers that have changed you know and so this guy Laurier develops this application you know to be able to facilitate those users you know informing themselves for those circumstances.
Now this is a bit of uh you know uh uh you know a corn it's not a total corner case you know but for dramatic effect there's more like you know use cases for this for you know more mundane planning purposes but it does speak to like when and why these things might be important right and so this thing starts its life not last year not the year before 2019 precoid like this story has been a long time in the making right and it starts as many things do as a classic skunk works project, right?
Hey, I'm going to like, you know, test out this theory. I'm going to pull some data from some smart meters, you know, I'm going to give it to a few like early adopters and we're going to see what happens, right? You know, and then remember this person is not a full stack developer, so he's not speaking full stack JavaScript, React, blah blah blah. Now, he knows enough Python to be dangerous. So, Python for him is a programming language, not a snake, right? you know, but it's not like the top part of his, you know, rais to use, you know, a bit of.
And actually, the UI here is on the right. It's a static image, so it doesn't really do it justice, right? You know, like an application, if you take a picture of it, you know, chances are, you know, if I took a poll of you and we hadn't informed and I asked you what this was, more than a few of you, you know, would call it a dashboard. And I go, "No, no, it's not a dashboard. It's not a dashboard." Right? application, right?
An application allows the user in there to do more than just view and consume data, but to actually like do what if analysis, you know, you know, calculate, you know, various things, do like iterations on that. And so here, uh, I did an animated GIF because I don't have access to the app. It's in a secure environment. I don't have privileges. It's a little bit grayed out, a little bit, you know, fuzzy. Even if it was fully clear, how many people here speak or read?
Oh, none. Well, I could have like done it in the clear, right? And you'd be okay. But basically, like this is all in. It's an application. It's a web application, right? And if you know, you know, the animated GIFs comes back and forth. But on the left, there's a whole bunch of filters. You can filter out like, you know, whether you want like everything up to 110 volts or 100 volts. So, you can narrow down your field of view.
And then from there you can select certain smart meters or transformers or subset of your grid. And then from there when you pull it you can pull it back and you can see a year's worth of trans of smart meter data, right? And then from there because you know what you're looking at you can see like macro variations. Oh like in the winter like you know it goes higher versus lower, you know, but then you can also see like you know you know little spikes and you know this and that, right?
And so like the person who knows what they're looking at you know they can do a lot with this thing, right? Right? So, it's a highv value application for that niche set of users, right? And just to reinforce, it's a data application, not a dashboard, right? You know, and oh, here's a little bit of uh, you know, advertising, right? How did he build it? Right? Well, he's in Python. He used this web application framework called Dash, you know, that's been developed and open source by a company called Plotley. that company called Plotley is based out of Montreal, you know, and he knew about this company back in 2019, right?
So, he was like an original gangster like he, you know, he was one of the early early adopters, you know, of Plotly Dash and he knew that he he could use this for this purpose. Now, Hydro Quebec, big company, right? Hydro Quebec, big Microsoft customer as are many big, you know, utilities, right? And so, Microsoft has many technologies, right? One of those technologies is PowerBI, right? PowerBI, you can do a lot of different things.
And because they're a big Microsoft, you know, customer, you know, there's a dictate, right? Thou shalt use PowerBI, right? Unless, right? And so he found the you know the corner case the unless hey I can use powerbi for a to t you know but uv wxy and z or zed depending on your procilities right you know not so much if at all and so he you know in python used this plotly dash web app framework because it afforded him these general kind of characteristics it was more flexible it was more interactive it was more scalable it was more customizable, it was more actionable and it provided more of an opportunity for it to be sophisticated.
So the expression of pain, you know, because he tried it first in PowerBI was I don't just want a dashboard anymore. I want an application. I want a UI. I want a tool and logical and yeah, he didn't want to hire a full stack dev team, right? Didn't want to go to Accenture. Could he do this thing himself? Right? And you know, this was 2019. And so when he did this, this is precoid and if you look at that chart on the left, that's actually a chart with a with a uh what do you call it like a logarithmic, you know, exponential uh yaxis.
So it shows the increase in use or adoption of this plotly dash web app framework over the course of time. So back then there were on the order of like 100,000 downloads per day. You know, roll the clock forward six years, five million plus, you know, per month, right? So lots you know lots of adoption in lots of countries. So actually I have the distinct privilege right in my job at plotti of waking up every morning talking to people all over the world doing all manner of crazy stuff.
This is one example of such right you know and I know this much to anything about what they do but they manifest themselves in these circumstances right hey we use X but status quo is not working but and so they tend to be the sophisticated advanced analytics use cases it's a prime example right so let's roll the clock forward it's not 2019 anymore it's 2023 right okay mission accomplished box office success, you know, this application called Octave.
And I asked like, you know, what why'd you call it Octave? And he said, well, it's some acronym that he concocted. He can't even remember why, right? But now he's on a version 2. 8. It's an application. You don't version control dashboards. Applications you do. So, he's actually got a regimented sequence like three and a half years later. He's gotten through COVID and now this skunk works project now has 350 users. Now remember the total population like the total headcount of Hydro Quebec about 22,000.
I think it's I think the number in the distribution wing is about 5,000. So you got 300 of 5,000 using this thing. It's not a bad thing for like a little skunk works project, right? And of those 350 upwards of 150 are using it per day. not for eight hours a day, you know, maybe for a few minutes, maybe for an hour, you know, and there's a fair number of queries and now they're accessing the entirety, right, you know, of the network, four and a half million smart meters and they have multiple use cases and they've actually operationalized it.
So, they actually have to have some like formalization regimentation and he's the only guy who's who's developing it. So, it's like, you know, it's a development team of one. So everything should be good, right? Wow. Actually, plot twist, right? Couple things, you know. One is that the scale on the back end, right? Now you got four and a half million. Now you got multiple users worth of data. The numbers start adding up, especially in aggregate.
And on top of that, because you're operational, there's more scrutiny. Oh. Oh, apparently, you know, smart meter data can be used to identify like what you're doing, right? Like if you're a, you know, a retirement home, like your actual peak loads and valleys are very regimented. Who knew, right? You know, and probably if you have a grow up, you know, there's some like characteristics, you know, that you could find out.
So greater scrutiny combined with, right, Quebec adopts a privacy law in 2021, 2022. It's a GDPR kind of equivalent. So you know thou shalt you know be standard. So for both scale and privacy governance reasons you know need to evolve you know the overall architecture. Well we're at a data bricks conference. What do you think the next slide's going to be? Right? You know you know enter stage right data bricks right? You know, and so they take all the data from the legacy, you know, whatever it was that was working reasonably well, which the end users don't care about, right?
And they pump it all into data bricks, right? And it addresses the data governance requirements, supports large volumes, you know, incorporates a really, you know, highly optimized pipeline. They got all these logical functions. They convert them to pi spark, you know, at each stage. They create these intermediate tables, and I'll get into that a little bit more. They partition, you know, they optimize, you know, through memory, their their clusters.
Life should be good, right? You know, they're doing all the best practices for this particular circumstance. So, what could be, you know, what could be wrong, right? Why we haven't Well, not so fast. Not so fast, right? The end users, they don't care about any of all that stuff going on underneath the hood. They don't care how the data gets to where they just want their data two seconds or less. It's a classic problem, right?
You know, and because if they don't, then they're going to be waiting. Where are they waiting? In a truck, you know, if they wait 15 seconds or whatnot, they're not going to use the application. So, it's going to go unused. But let's look at what's happening below the waterline. Lots happening, right? The one thing that I learned, you know, from an appreciation standpoint is that this topology changes daily, right? I alluded to it previously.
You know, new smart meters are getting, you know, plugged in, right? new transformers, other ones. So the dynamics of the network change all the time and it changes in context with lots more data. And so as they as the end users want to see the health of the network, they actually running you know um you know you know pretty intensive um you know computations you know to get the state of affairs right and they're doing it ideally daily but they're actually doing it weekly because it's super expensive right super expensive to do but it's despite its expense it's worth enough time and energy, you know, and expense for the end users to use it.
They wouldn't be spending the money, you know, on this on these updates if it wasn't worth it for the end users. And then they have these, you know, these these um profiles optimized. So, lots going on there. The more you ask about this, the more I'm going to have to defer you to a data bricks, you know, team, you know, or because even Laurier, who's the electrical engineer, he can understand the symptoms, had to talk to the data bricks people, you know, to facilitate this, right?
So the challenges that persist, one is the fact that despite some of these optimizations that they did from a standard standpoint, the interactions were slow and they're compounded by the fact that there's more users, right? More users who are using this thing concurrently with more data. So if you look at, you know, if you if you train your eye on the right, you know, this is actually like a seven, eight second lag, right? you know, and so it might be okay once, but if every time you got to like wait seven seconds, you know, it becomes a disjointed experience, right?
So he's trying to like reduce the amount of time lag, you know, so end users can have this interactive, you know, graceful experience. So they do a number of different things in terms of like, you know, weekly updates on various clusters, you know, they they add certain like workers from their, you know, from their pool. Um, and so this that's around 2023 after they've adopted. And so when these challenges persist, you know, this is where they go to the domain experts, right?
And this guy goes to his, you know, his his mafia, you know, his data bricks account team in Quebec who treat him really well, right? And they've got really good data engineers, right? And they get bring in the specialist, right? and they work and they basically understand what the what the real world dynamics are, what he needs from a business standpoint to achieve his technical and business objectives, right? Ultimately to reduce query latency, better concurrent usage, you know, but he had to do that all in the context of satisfying the underlying like ETL processes, right?
And so you know they actually did a fair bit of code you know analysis of the not just the app you know but the underlying uh you know sequence you know in terms of optimizing. So it's a team effort you know that's kind of one of the takeaways here like we're not you know in a vacuum you've got your five different you know customers right you know if you have a center of excellence you can bring these learnings you know to the table and in fact has been very very generous right in terms of you know providing opportunity to speak to other utilities with similar symptoms right we talked to a utility in Sweden just a couple weeks ago you know because they've expressed similar symptoms now there's less obvious solutions had any data engineers sitting in the audience.
All right, so you're going to understand this better than me, right? So you you know I can send you the PowerPoint, you know, you can like read, you know, read through this, you know, but these were other steps, right? I don't profess to understand this. I actually asked Laurier and he like deferred the question, right? So I bring it up for your benefit, right? you know and if you want to know more I'm happy to introduce you to people who can like speak to the you know to the nuance of this you ultimately though at the end of the day right the success factors this is not an overnight success right like this is a this is actually in some ways a victim of its own success right the skunk works project it's classic tale right you know and then all of a sudden it's like bailing wire and duct tape up to a certain point you know but becomes so ingrained by these users that now you actually have to make it real, right?
And so this has shown that set of techn that combination of technologies can take you to that destination, right? And so like that's an important point to note. But you nothing replaces domain knowledge, right? Another key point in this is autonomy, right? Under underappreciated underappreciated especially in big companies, right? He's in a special team, right? He's in an advanced analytics team. he's afforded the opportunity to be commissioned on these types of projects.
If he wasn't, right, you know, then you have the classic, oh, we got to give it to the full stack dev team, you know, and they're like super smart, super smart, full stack developers, right? But they're not domain people. So there's a lot of lost in translation, right? And we're not talking English to, to English. We're talking like domain speak, you know, to compsize speak and back and forth, you know, and you know, I think we've all experienced that, right?
Another one is code quality. We're not talking vibe coding here. We're not talking a quick dashboard. This thing has a life, right? And so he despite this small team is actually like a miniature, you know, application development team, right? So you got to adhere to some protocol and you know standards. You got to document your, you know, your code, do some unit testing, right? And then rely on the people who know, you know, what they're doing, right? data bricks you know provided like a lot of you know lot of technical competence you know because they're you know they've seen some of these things in different ways shapes and forms you know globally right and you know from our standpoint here at plotley right he's an electrical engineer we see web applications data applications like that's what we do right that's what our people do so whether it be like UI UX you know like an an application doesn't get used if it doesn't look good.
It just doesn't right. An application doesn't get used, you know, if the user experience is not good. You know, and in fact, an application use will be detracted like people won't trust the math if it doesn't look good and if you know, and if the user experience is bad, it'll reflect poorly on the actually underlying analytics. So, you know, we we've had the opportunity to help him. Just make sure like you finesse the details.
So, it's a franchise in the making. 2024 2025 as of May. So, a month ago, thousand registered users. This guy's continued to develop the application amongst other things. He's got other projects. This is not his only like iron in the fire, right? He's actually advancing other use cases where you saw that map, right? that map that had, you know, transformers and it had smart meters. Well, he's actually doing some interesting like network geospatial, you know, topology, you know, mapping where we can like move the, you know, move the placement of the transformers around, rerun calculations based on, you know, the distance and the wiring and everything else, right?
And, you know, get that into the use of these end users, stuff that in their regular tooling, you know, can't do. And then there's like other things. Now, Hydro Quebec is a big company. big companies move slowly, right? So these, you know, so they're just getting Unity catalog into place and once they do, they will be able to afford themselves a more like recent advances with serverless. But serverless may or may not, you know, provide like, you know, a Nirvana type of, you know, thing because they have so many users all the time that they have a dedicated cluster, it's up and running.
You know, it might be a six of one, half dozen of the other, you know, type of scenario. there's other things that they're looking at, right? You know, and then some of it becomes a point of like point of diminishing returns, right? You know, is it worth the cost, right? Because at this point, the users, they're, you know, they're happy, they're satisfied, right? And at that point, that's, you know, we're at a good enough stage.
So, you don't need to overengineer the thing at the same time. All right, that's the credits, right? So, we've been here two days. If you happen to be here tomorrow, we're at the booth F612. If you want some more PlotL dash things, well, you can't see them because they were yesterday, but those three uh you know, those three uh use cases were all like financial services oriented. The last two were actually about real time.
So, this you know scenario it was you didn't see like things updating on the screen every second, right? So it's it's it's a historical and smart meters you know for those you know they don't they don't pump out data hertz frequencies you know it's more like five 10 15 minute intervals right so you're not looking like you know did it change in the last two seconds but you can use these web applications in Python with dash with data bricks for these of things and as you you've seen in some of the keynotes even just today right this is where you get into like really sophisticated stuff with OLTP versus OLAP and you know bringing using Postgress and we get to see and experience those types of scenarios.
Want any other information? There's lots of different other use cases that are available. Industrial IoT, energy and utilities is like between that and financial services is probably like the two most common uh like intersection points, right? Where we see our technology and data bricks used, you know, very well together. And um you know with that I got five minutes. You got any questions? I got you know a little bit time thereafter and oh yeah not everything works right so there's you know there we tried you know they tried other different things you know and uh not everything worked in that's case all right well thank you for your time thanks for sticking through at six o'clock

Learn more about the Databricks Data and AI platform.