What is Edge Computing for Data & AI, and Should You Be Interested?

Summary

Ben Dunmeyer, a data analytics consultant at Thorogood, discusses the impact of Edge Computing on data and AI. He explains the concept of processing data near its source to reduce latency, enhancing real-time decision-making. Dunmeyer explores use cases like retail inventory tracking, workplace safety, and healthcare monitoring. He also delves into the practical challenges of implementing Edge Computing, such as hardware requirements, security, and aligning with existing data strategies. The talk concludes with thoughts on the evolving landscape of providers and the importance of integrating Edge Computing into long-term data strategies.

Takeaways

• 🌐 Edge Computing brings computation and storage capabilities closer to where data is generated, reducing reliance on central data hubs.
• 📈 It enables low latency or real-time data processing and analysis, which is crucial for applications requiring instantaneous decision-making.
• 🏭 The technology is particularly impactful in industries like manufacturing, healthcare, and retail, where real-time data can optimize operations and enhance safety.
• 🔧 Efficient use of physical assets is achieved by utilizing edge devices to perform computations and analyses during idle times, maximizing ROI on infrastructure.
• 🆕 The ability to combine new and diverse datasets collected through edge devices can lead to novel insights and improved decision-making across various sectors.
• 🛠️ Edge Computing allows for the filtering and prioritization of information, reducing the need to transfer all data back to a central hub and thus optimizing network usage.
• 🌟 It can operate effectively in environments with limited or unreliable network connectivity, making it suitable for remote locations or situations where network infrastructure is a constraint.
• 🛡️ Security and data sovereignty become critical considerations as sensitive data and intellectual property may be stored and processed on edge devices.
• 🔄 The technology's potential is heavily influenced by the capabilities of cloud platforms, which provide the infrastructure and tools necessary for edge computing implementations.
• 🔧 Successful integration of edge computing requires a strategic approach that considers technical capabilities, business objectives, and the physical realities of deploying and maintaining edge devices.

Q & A

What is the main focus of the discussion in the transcript?

The main focus of the discussion is Edge Computing and its implications on data and AI, particularly looking at its potential impact on the data and AI landscape.

What is Edge Computing?

Edge Computing refers to the ability to have both compute and storage within devices or where data is being generated, as opposed to relying solely on a central data hub or infrastructure.

What is the significance of low latency in Edge Computing?

Low latency in Edge Computing is significant because it allows for real-time or near real-time data processing and analysis, which is crucial for applications that require instantaneous decision-making.

How does Edge Computing relate to AI applications?

Edge Computing can enable AI applications by providing the necessary computation and storage capabilities close to where data is generated, allowing for faster and more efficient AI model training and inference.

What is the role of Thorogood in the context of this discussion?

Thorogood is an independent Global Professional Services firm specializing in business intelligence and analytics strategies, solutions, and services. They offer a full range of data analytics services and work with various technologies to provide clients with solutions that best suit their needs.

What are some potential use cases for Edge Computing mentioned in the transcript?

Some potential use cases for Edge Computing mentioned include real-time decision making, efficient usage of physical assets, combining new and diverse data sets, filtering information, and reducing dependency on environmental or physical limitations.

How does Edge Computing impact traditional business intelligence reporting?

Edge Computing can impact traditional business intelligence reporting by allowing for real-time or near real-time analytics, which can provide more up-to-date insights and potentially lead to better decision-making.

What are the challenges associated with implementing Edge Computing?

Challenges associated with implementing Edge Computing include physical practicalities such as procuring, installing, and maintaining the equipment, ensuring connectivity, addressing security concerns, and considering data sovereignty.

How does Edge Computing fit into an organization's overall data and AI strategy?

Edge Computing should be considered as a facet of an organization's overall data and AI strategy, where it can be opportunistic for specific use cases or part of a wider rollout. It's important to align it with business and technical strategies and consider how it will work with existing systems.

What are some of the core capabilities needed to operate and manage Edge Computing?

Core capabilities needed to operate and manage Edge Computing include operating and running systems, data systems, devops, data engineering, data science, machine learning, data collection, monitoring, and potentially mlops (machine learning operations).

How might the landscape of providers for Edge Computing look like in the future?

The landscape of providers for Edge Computing may include the mainstays of the data and AI space, new players, conglomerates, or partnerships. It will be interesting to see how the hardware requirements and market potential shape the competition.

Outlines

00:00

🌐 Introduction to Edge Computing and Its Impact on Data and AI

The speaker, Ben Dunmeyer, a data analytics consultant at Thorogood, introduces the topic of Edge Computing and its potential implications on data and AI. He outlines the agenda, which includes sharing insights into Edge Computing, its current implementations, and its future impact on data and AI landscapes. Ben emphasizes the importance of understanding Edge Computing, particularly in the context of low latency data processing and analysis, and how it can enable real-time decision making and efficient use of physical assets. The discussion also touches on the need for filtering and prioritizing data to optimize business intelligence and reporting.

05:03

🔧 Edge Computing in Practice: Real-World Applications and Opportunities

This section delves into practical applications of Edge Computing, such as enhancing workplace safety, optimizing retail operations through inventory tracking and promotion analysis, and improving manufacturing processes. The speaker highlights the ability of Edge Computing to collect new and diverse data sets, which can lead to fresh insights and improved decision-making. It also addresses the potential to reduce dependency on environmental factors, such as limited network connectivity in remote locations, by processing data locally. The paragraph provides various examples to illustrate the versatility of Edge Computing across different industries.

10:05

🏥 Emerging Use Cases of Edge Computing in Healthcare, Transportation, and More

The speaker explores emerging use cases for Edge Computing, including patient monitoring in healthcare, traffic management, and autonomous vehicles in transportation. These examples showcase the potential for Edge Computing to enhance real-time responses and decision-making in various sectors. The discussion also considers the future growth of Edge Computing as technology advances and the volume of capturable data increases. The speaker invites the audience to consider how these use cases might apply to their own organizations or industries and to think about the potential long-term strategy involving Edge Computing.

15:07

🛠️ Implementing Edge Computing: Technical and Strategic Considerations

In this part, the speaker discusses the implementation of Edge Computing, focusing on integrating edge devices with existing data and AI architectures. He uses a manufacturing plant as an example to illustrate how data collected by edge devices can be used to monitor equipment, identify potential faults, and improve quality control. The speaker emphasizes the need for a robust data platform to collect, transform, and analyze data from edge devices effectively. He also touches on the importance of retraining models and managing operations on edge devices, suggesting that successful implementation will require a combination of data engineering, data science, and DevOps practices.

20:09

🌉 The Future of Edge Computing: Challenges, Strategies, and Provider Landscape

The final paragraph addresses the challenges and strategies associated with Edge Computing, such as physical practicalities, security concerns, and data sovereignty. The speaker also considers the fit of Edge Computing within existing data and AI strategies and the potential for it to be either opportunistic or a core part of a broader technological rollout. The discussion includes the evolving landscape of providers and the possibility of new players entering the market. The speaker concludes by encouraging the audience to think about how Edge Computing might fit into their long-term data strategies and to consider the practical steps for implementation.

Keywords

💡Edge Computing

Edge Computing refers to the concept of performing computation and data processing near the source of the data, rather than in a centralized data processing warehouse. In the video, it is discussed as a transformative approach that has significant implications for data and AI applications, particularly in reducing latency and enabling real-time decision-making. The script mentions how edge computing can be applied in various scenarios such as retail, manufacturing, and healthcare, showcasing its broad applicability.

💡Data AI

Data AI, or Artificial Intelligence in the context of data, involves the use of AI techniques to analyze, interpret, and derive insights from data. The video script discusses how edge computing can impact the data AI landscape by allowing for more efficient data processing and analysis at the edge of the network. This can lead to new opportunities in real-time analytics and decision-making, as highlighted by the speaker's examples of using edge computing for instant decision-making and efficient use of physical assets.

💡Latency

Latency in the context of computing refers to the delay before a transfer of data begins following an instruction for data transmission. The video emphasizes the importance of low latency or no latency capabilities in edge computing, which is crucial for real-time data processing and analysis. This is particularly relevant for applications like autonomous vehicles or industrial equipment, where immediate response times are critical.

💡Thorogood

Thorogood is mentioned as an independent Global Professional Services firm specializing in business intelligence and analytics. The company is characterized by its consultants' unique blend of business understanding, industry experience, and technical expertise. In the video, the speaker from Thorogood is discussing edge computing, indicating the firm's involvement in advising and implementing edge computing solutions for clients across various industries.

💡Real-time Analytics

Real-time analytics is the process of analyzing data as it is collected, without any significant delay. The video script discusses the potential of edge computing to enable real-time or near-real-time analytics by performing data processing close to the data source. This capability is highlighted as a key advantage for applications that require immediate insights and actions, such as monitoring equipment in a manufacturing plant or tracking inventory in retail.

💡Data Sovereignty

Data sovereignty is the concept where data is subject to the laws and regulations of the country in which it is collected, stored, and processed. In the video, data sovereignty is mentioned as a potential challenge in the context of edge computing, especially when data is generated and processed across different geographical locations. The script implies that companies need to consider data sovereignty when implementing edge computing solutions to ensure compliance with local data protection laws.

💡DevOps

DevOps is a set of practices that combines software development (Dev) and IT operations (Ops) to shorten the systems development life cycle and provide continuous delivery of high-quality software. The video script mentions DevOps in the context of managing edge computing infrastructure, suggesting that having a robust DevOps practice is crucial for the successful implementation and management of edge computing solutions.

💡MLOps

MLOps refers to the practices, processes, and tools that enable the deployment, monitoring, and maintenance of machine learning models in production. In the video, MLOps is discussed as an essential capability for managing machine learning models deployed on edge devices. The script implies that MLOps practices are necessary to ensure that models on edge devices are up-to-date, performant, and reliable.

💡IoT (Internet of Things)

IoT refers to the interconnection of various computing devices with sensors, software, and network connectivity, allowing these devices to collect and exchange data. The video script discusses IoT in relation to edge computing, highlighting how IoT devices can generate data that is processed and analyzed at the edge of the network. Examples given include sensors in a factory or a smart building, illustrating how IoT devices can benefit from edge computing's low-latency data processing capabilities.

💡Data Lake

A data lake is a system or repository of data stored in its natural format, usually a single store of all enterprise data including raw copies of source system data and processed data used for tasks such as reporting, visualization, and analysis. In the video, the concept of a data lake is mentioned as part of the data architecture where edge-computed data can be landed and further processed. The script suggests that data lakes can play a role in storing and organizing the vast amounts of data generated by edge devices.

Highlights

Introduction to Edge Computing and its implications on data and AI.

Definition of Edge Computing as the ability to have compute and storage within devices where data is generated.

Discussion on the potential for low latency or no latency capabilities in data processing and AI applications.

Introduction of Thorogood as an independent Global Professional Services firm specializing in business intelligence and analytics.

Explanation of how Edge Computing can enable real-time or near real-time analytics.

The importance of efficient usage of physical assets and how Edge Computing can optimize it.

Opportunities for combining new and diverse datasets with Edge Computing to uncover new insights.

The ability to prioritize and filter information with Edge Computing to reduce unnecessary data transfer.

How Edge Computing can reduce dependency on environmental or physical limitations.

Examples of Edge Computing in retail, manufacturing, workplace safety, and healthcare.

The intersection of Edge Computing with data engineering, data science, and MLOps.

The role of cloud vendors in providing infrastructure and software for Edge Computing.

Challenges in implementing Edge Computing, including physical practicalities and security concerns.

Strategic considerations for integrating Edge Computing into existing data and AI strategies.

The future landscape of providers in the Edge Computing space and potential partnerships.

Practical steps for getting started with Edge Computing and aligning it with long-term data strategies.

Transcripts

foreign

[Music]

thanks everybody for coming along today

we're going to be talking about Edge

Computing and particularly looking at

its implications on data Ai and really

just thinking about where this kind of

functionality capability could go pretty

simple agenda really aiming to share a

little bit about Edge Computing kind of

understanding of it where we're starting

to see some like sorts of

implementations with customers or at

least talking about it and gives kind of

exposition as to like how we think it

could and probably will impact the data

and AI landscape just quickly to

introduce myself and then Thorogood so

my name is Ben dunmeyerman data

analytics consultant I'm based in the US

so for those of you that don't know

Thorogood we're an independent Global

Professional Services firm specializing

in business intelligence and analytics

strategies Solutions and services or

global company with offices in Asia

Europe and the Americas as I mentioned

I'm in Philadelphia all our Consultants

are recruited and trained in the same

way to develop a unique mix of skills

blending business understanding in the

form of industry and functional

experience with a strong technical

aptitude and deep understanding of

analytical tools and techniques and we

do offer a full range of data analytics

Services you can see some of those on

screen in terms of the technical

Technologies we work with we are an

independent consulting firm meaning we

don't work with one specific technology

but really work across the best in breed

tools and platforms across the industry

so that's big players as well as Niche

providers and really what we want to do

is partner with with these key players

in the market to provide our clients

with a solution that best suits their

needs and as we talk about Edge

Computing there's probably Technologies

and vendors that are on the screen but

will evolve into the the future but

we'll keep it in context so starting out

just giving an idea and maybe a little

bit of a definition of edge Computing

and where we see the intersections with

particularly data and AI applications

won't really get into things like

Hardware or that's configurations so

again starting simply grabbed a few

things from the public domain to kind of

present the topic and I guess the

concept of edge Computing is relatively

Sim simple with just like anything that

and kind of more than meets the aisle a

lot to actually implement it but the

fundamental idea is that within devices

and the technology advances we have the

ability to have both compute and storage

within devices or where data is being

produced or generated as opposed to just

in some sort of central data Hub or

Central kind of firewalled

infrastructure so the idea of a sensor

in an office that is gauging temperature

number of people in a room or an

autonomous vehicle or piece of equipment

in a factory they're all kind of

mechanisms or Machinery or devices that

can generate data so that idea of edge

Computing is bringing the ability to

have some level of computation some

level of storage kind of on or right

next to that device a big idea is being

able to achieve a low latency or really

no latency capabilities with data

processing data analysis and and

potentially AI applications to see maybe

the diagram on the right kind of draw

always had a couple number of different

topics some examples of these sorts of

devices and it's really thinking about

like what do I want to accomplish and

what are the latency requirements for it

and and thinking about the kind of

business implication of it is there a an

opportunity or a need for Real Time near

real time or actual real-time analytics

or is it that and also some some needs

for kind of traditional business

intelligence reporting or operational

reporting so kind of in principle it's

introducing this concept and it's been

around for for a number of years I think

getting more and more popular and

practical

um it's putting the the technology and

capabilities in place where data is

being produced so we can do potentially

some of the same sorts of operations or

analyzes that we do now in a you know

centralized data store centralized data

Hub that we curate and build really I

guess I want to think about like how we

might apply these sorts of scenarios

what scenarios they might apply AI for

and what the value add could be so here

I've sort of laid that out um kind of

what is the opportunity and then how

does Edge computer Computing sort of fit

that bill or allow us to do it so the

first one I mentioned this is now a

couple times is real time or

instantaneous decision making as so

really and again kind of hits the nail

on the head with the the ability to have

some level of computation and storage on

the device or right next to the device

so to speak that doesn't require a

movement through networks to some sort

of data store and other batch or

streaming processing so effectively

allowing us to do low latency or really

maybe no latency analysis as the data is

being produced efficient usage of

physical assets so what I really mean by

this if we I guess think about

Technology and Manufacturing sites or

smart buildings or kind of any number of

other kind of physical things that kind

of Hardware if we use that term pretty

Loosely is like going to be I title at

times there's not always going to be

someone in a roommate given piece of

equipment is not always going to be

running but if it has these capabilities

of edge compute within it or as part of

its structure framework how do you kind

of use them to best affect

um like if they're there and they're

just sitting idle we want to be using

them maybe running some analyzes or

retraining a model or something like

that so it's it's I guess the

opportunity is getting the most out of

Investments for their physical

Investments or data Investments and it

like thinking about how we can use those

Investments all the time and even when

they're idle it's just something that's

I think near and dear to us in the data

analytics Community is the ability to

combine new and diverse data sets to get

to New Perspectives or come up with new

analyzes or uncover insights that have

previously we haven't thought about or

hadn't been able to prove or disprove

and a similar idea here with Edge

Computing I think the the angle with

Edge Computing is is the kind of nascent

or new data sets that can be acquired if

able to tap into this capability and a

lot of this is is probably technological

advances in building these components or

equipment that is able to capture

information whether it's like Biometrics

or information about a vehicle or again

in a plant a couple of these common

examples I've been using

so like with the opportunity to collect

new data what new things can we come up

with can we improve decision making in

like in any number of areas um and I'll

go through some examples later so the

opportunity to collect new and diverse

data that we can use in like lots of

different ways the ability to kind of

prioritize our filter information so we

think about any sort of routine you want

to know should I be doing this should

this data that I'm collecting from X Y

or Z do I need to bring it back into my

central Hub do I need to store it do I

need to process it or should that

investment or that time be spent

somewhere else so with with the idea of

compute kind of where the data is being

collected we might be able to make

decisions and to exclude erroneous data

or do some like initial evaluation as to

what information like has any purpose

versus what things we you know don't

store do store so I think it gives us a

little bit more autonomy and and maybe

auditability to decide for things like

Internet of Things sensors what

information do we want to bring back

into a central data server because

that's going to be an investment we have

to set up those kind of physical

connectivity if to have the routines

that do it they have to be maintained so

can you filter out unnecessary

information for example in the last

example I have here kind of thing to

call out is the ability to reduce

dependency on any sort of environmental

or physical limitations or factors and

this one's a little bit severe probably

in some instances there's probably just

a few actual scenarios that come to mind

but if we think about an oil rig or

remote locations across the world or

somewhere in the ocean like an oil rig

not necessarily have that high speed

Network you might have you know High

latency Network or a little bandwidth

very likely so the average Computing

giving us some opportunity to do some

computation some data storage where we

don't have the ability to pull all that

information or a lot of that information

into some sort of central place

and this one's pretty severe it's

probably a few instances of it but I

think it's it's pretty interesting so

I've alluded to a few and I think

there's some kind of obvious examples of

this and I'd only expect this to grow

over time as just like technology

advances uh Hardware advances and our

the amount of information we capture

just becomes and people have seen

studies exponentially grows so a few

examples here kind of broken them into

some some groups can maybe briefly hit

on a few of them or maybe all of them

just because they get some interesting

examples and I might be thinking about

where it applies to your organization or

the industry that you play in

so starting from the left retail

scenario where I've done a lot of of

work and a lot of my expertises and so I

think retail both on the front and the

store and in the the manufacturing side

which is just to the right but things

like stocking tracking stocks so where

are we low on inventory whether that's

in the warehouse or or in the actual

brick and mortar stores potentially the

ability to maybe not in the moment but

pretty quickly get information about

promotion Effectiveness so if we run a

promotion in a certain display or in a

certain aisle is the foot traffic in

that aisle the days or times when the

permission is active or they increase or

they decrease or stay the same and like

the ability to see that and potentially

make almost real-time decisions is

pretty wildly interesting uh workplace

safety so you can think about that

across various different construction

zones and any other I guess

manufacturing the top here I made I said

the example earlier about it like oil

rig but sorts of isolated areas again

this is a pretty severe example because

there's not that many but how can how

can these scenarios be like better

managed or how can we use data analytics

in the scenario where we don't have the

ability to pass a high amount of data to

a central data store and get it back and

make any use out of it Healthcare so you

might think about patient monitoring so

being able to pick up on the numerous

biometric and other signals that you can

get from monitoring a patient something

like room availability in a hospital or

an office Transportation so autonomous

vehicles one that is kind of up into the

news right now traffic management

monitoring emergency vehicle responses

or even Pathways so transportation is an

interesting one that one that a lot of

us experience every day and a few other

words here I will probably leave them to

to read through but I think number of

interesting use cases to think about it

as I said I think it will only grow as

technology advances as we're able to

collect this data from more diverse

places and it's kind of interesting to

see where it goes and ultimately if the

hardware and technology is in place to

be able to do it can we actually take

advantage of it so I'll try to frame us

a little bit in how the kind of concepts

of educated Computing and some of these

use cases which are kind of a little bit

ubiquitous or generalized how might it

impact data and again not going to jump

into any details here of how to do it

but I really wanted to Think Through

like what are the components and how

what may adopt or use the theories and

the practices that we've developed over

years and continue to develop and kind

of building data architectures and these

use cases which present some physical

limitations and like hardware and what

we can do I'm in some kind of practical

limitations I suppose Give an example

and again not to any level of detail but

but how hopefully frames a little bit

and gets us thinking about like what we

need to make this a reality or potential

reality so I'll use the example of a

manufacturing plan I think it's a pretty

on the nose example just given the

number of different use cases like

monitoring equipment for controlling it

but also for identifying things that

have a high probability of failure or

faults given amount of time or usage you

know pretty common use cases other

things being like climate control or

building control which is just not

necessarily manufacturing identifying

quality issues so if you can spot

problems with Goods ahead of time using

some sort of video surveillance or even

information coming from the Machinery

that is producing the products we can

get ahead of quality issues and I'm sure

an endless list of theories and ideas

but maybe let's for the example think of

a classification or probability model

that we've developed and want to apply

in order to identify faults or failures

or the likelihood of them on either

subset equipment or certain you know

part of the process whatever it might be

so we have that and we have our Edge

devices across this manufacturing plant

and and we of course have our Enterprise

data and eye architecture doesn't really

matter the technology stack likely

multi-cloud multifaceted First Step but

luckily we are bringing this data into

our data platform so I've assumed some

sort of data like here and again this

I'll get to the actual Edge part of it

but with the edge device or the internet

of thing or things device we're able to

pass through Telemetry and other

information in some sort of structured

way and we'd land that data so whether

it's files or presumably in files we

landed somewhere in a data Lake and then

we have a pretty kind of natural way or

or learn way of curating that data

whether it's streamed or batch so we

might organize the data apply

restatements do some data cleaning

whatever else it might be that that

makes data either more presentable or

easier to interpret or whatever might be

often kind of interpreting logs is

something that we run into we know how

to do that we do that all the time

that's the bread and butter of a data

and AI architecture collecting data

transforming it and then using it for

various different things so it's the

same more or less here we're not

thinking about like a high velocity of

data and I'll kind of form it with the

picture and come back to it so with that

curated data we can do some data

analysis we'll do a reporting we do

analytics on it whatever else so we've

we've kind of built something that we're

pretty familiar with and we'll assume

that we're not necessarily feeding a

stream of data and we don't need to do

that in this case I think this could be

a batch or it could be every so often

collected from the The Edge devices

where we have the facility so the kind

of the scalability and elasticity of

cloud and and the kind of software to do

so we'll build and train a model so

again we have history data we have

whatever hours and hours or months and

months of telemetry data from these Edge

devices and we're able to develop a

model that is helping us to predict

either the probability or or the

likelihood of a fault or failure for a

piece of given equipment so again I

wouldn't expect us to be building or

trading this model I guess you could

train it on the edge device but you're

first using that device to collect

information maybe to do some potential

structuring of it but but then

developing a model and then deploying

that model so having some mechanism to

physically push that model into or onto

the edge device kind of completing the

feedback loop so to speak and the idea

here really is we're taking advantage of

our known cloud and data architecture

where we can build these sorts of

solutions and able to then kind of move

some of the compute or storage

requirements onto the edge device so we

don't need this data to come in at a

super high velocity all the time into

our kind of main platform we're able to

have a model on the device that

interprets this data and whatever spits

out a result on a gauge or a user

imagination exactly how it transmits

this information and then the whole idea

of kind of retraining so we're at

whatever intervals sending new Telemetry

data back into this platform or maybe we

have some ability given the compute

possibilities on the edge device to do

some kind of example retraining or to do

some monitoring and log and detection of

any skew or detection of any drift of

the model for example on the edge device

and then we take that information along

with new data send it back through our

kind of loop retrain a model or whatever

it is and then redeploy it practically

we're not going to do everything on an

edge device but we're we're taking

advantage of a number of different

facets so doing a bit more on that

device and integrating it with our kind

of tried and true platform the other

things that really stick out to me are

some of our core capabilities for

operating and running systems Data

Systems and if we think about doing

something like this if we're relying on

edge devices to run a manufacturing

plant you've got safety to consider you

are you're the livelihood of the

businesses is predicated on this you

want to have some reliable consistent

routines that allow you to manage this

so introduce the Ops you can put

different acronyms in front of it but

devops and envelops particularly in

these instances so do you have the right

devops set up to be able to manage the

data engineering and data collection

routines the moaps configuration to be

able to manage the model to be able to

interpret information coming from the

edge device to understand drift

understand log information collect

collect the data and retrain the model

or just rescore the model and and then

what what does it actually look like on

the edge device how are we deploying

code more or less to that edge device

I'm showing practical ways to to do that

but what does it look like there how are

we you know introducing some basic

routines to either clean data or produce

an analysis or even produce an output

that then gets visualized in some sort

of screen at the manufacturing Planner

on the device or whatever it might be so

I'm starting to think about not just

like the data manipulation the data

Transformations the routines but how do

we actually can manage it and again

we'll we'll try to answer it here it

won't get into any details but kind of

where my head goes looking at the tool

sets now and I think this is an area

where we're going to see the most

development the the most uh growth

really always not always start but

naturally start with the big cloud

vendors and and usually the leaders in

the space particularly for data and AI

so Azure AWS and and gcp a few other

players out there as well but the the

big players I've just listed out here

the the technology Stacks if that's the

right word that it seems that's how the

the software vendors are positioning

them as kind of stacks of edge or stacks

of iot capabilities and and again just

list of the names here for for awareness

really really what it looks like these

the cloud players are doing is they're

figuring out the infrastructure side of

things and that'll be interesting play

does does Microsoft produce devices for

for these different use cases this AWS

gcp do the same sort of thing or are

they going to create the software that

can get installed on edge devices which

are created by traditional manufacturers

so it'd be really interesting to see

certainly across each of these platforms

and the build out of the technology

stack and and really it's a an

infrastructure capabilities and then

leaning into the data analysis machine

learning data collection monitoring

capabilities that will needed to

ultimately kind of see a edge Computing

Vision come to life some of the kind of

questions that come to my mind things I

was asking myself list them out here do

we see the big Hardware producers Dell

IBM others playing in this place or they

are they you know they're expertise in

Hardware are they going to develop the

hardware the actual Edge devices partner

with the cloud underwear or try to do it

themselves do we see digital native

companies innately able to to do this

and in that case it may not we may not

be thinking physically as much but are

they they able to either produce maybe

produce products that we're adopting and

using these scenarios fully is it kind

of fully packaged together if we think

about those that are actually

manufacturing this equipment so building

sensors or building the edge devices

that go and sit on Machinery or a

turbine or whatever it is are they

partnering with one of the cloud vendors

are they doing it themselves so just I

think it's interesting to see I don't

think we quite know a lot will probably

depend on the how lucrative this this

space is who jumps into it who's able to

make it a practicality and I think the

world we live in that right now with

microchip and like precious metal

challenges and supply chain does that

slow this down or does that construct

the the players that could potentially

compete here so this is I think an area

we have to we have to monitor it to see

what the cloud software data vendors are

doing and see who else either enters the

space the other takeaway I have is is I

bet it's not so different but but it's

not so different than what we've always

had to do it's just happening in a

different place and I suppose introduce

producing some physical practicalities

and some the need to be responsible in

the usage of The Limited compute or

limited storage that we have but really

it's that intersection of data

engineering data science visualization

devops mlops and probably other bubbles

you could put on here but it's it's

getting all these right into a single

solution to make it a a practicality you

know if we have inefficient routines

that collect data and transform it we're

not going to actually take advantage of

that low latency with no latency

connection if we have routines that take

forever to run

um if we don't have the appropriate ml

or devops capabilities to kind of make

changes in in real time or collect

enough information to adjust

infrastructure configurations or

whatever it might be it's not actually

going to to work so we've got the same

foundations of data and AI existing here

but it's obviously learning about the

nuances or the the keys to make

something like an edge Computing use

case or Edge Computing implementation

successful and also lay is what are my

basis objectives what's my strategy my

technical strategy my business industry

strategy and with someone's physical

footprint there's there's a obvious

capital expenditure here do you retrofit

existing facilities or equipment with

these devices is it only net new you

know smart buildings or smart grids or

new vehicles that you apply these

sensors to so there's questions there as

to like what's the investment where does

this apply and how do you just like we

always are thinking about the merging of

Legacy systems with new systems which

just is a continuous cycle how is it how

does it work here when um pretty diverse

I'm a kind of diverse new application

so I'll close that here just about out

of time and just leave a few thoughts

here I didn't really talk about the

challenges I guess some of them are are

obvious but like how do you come up with

some approaches that that do combat the

kind of known challenges Edge Computing

and and the ones that we kind of expect

to to see as as it actually becomes a

reality so there's the physical

practicalities like I mentioned of

actually procuring the equipment

installing it maintaining it over time

you do have to have some level of

connectivity what about security uh so

if you just kind of have data and

potentially some intellectual property

stored on these devices do you need to

secure those devices I don't know what

to think about data sovereignty at all

probably what's the fit with your

existing data AI strategy I mentioned

this on the previous slide or two is

like where does your business want to go

where you're going with your technology

strategy and and how do they work

together again I think Edge Computing is

a facet of one's overall data AI

strategies it could be opportunistic of

use cases it could be a kind of wide

rollout and a big bet that someone makes

but I think if you you can get that

right if you can figure out the

scalability of kind of a base Cloud

architecture that is cloud does enable

us to do if you can get that right with

Edge Computing with these other new and

interesting capabilities which are you

know next year it'll be something

different like if you're able to scale

those together and effectively and with

minimal rework I think that's the real

differentiator I mentioned a little bit

what's the what's the landscape in terms

of providers look like are we going to

see the the Mainstays and the kind of

data and AI space like lead here will be

new players will be kind of

conglomerates or Partnerships it'll be

interesting to to see given the the

uniqueness potentially of the of the

hardware that's needed and then really

how would you get started or I'm sure in

certain industries it lends itself

naturally to that are already doing this

and how do you accelerate how do you

make better use of the investment how do

you kind of make use of idle time all

those sorts of things and and again

maybe impossible question to answer but

an area to start to start thinking about

either use cases you have or opportunity

communities that you see to implement

this or an assessment of what your 10 15

20-year data strategy might look like so

I'll close out here I just listed out a

few areas that they're a good can help

and again and kind of some of the

exploration or thinking around Edge

Computing but a number of other things

and you can see those listed here kind

of a full range of services that we

offer of course do do get in touch if

you want to talk talk about this topic

or any other topics always happy to

discuss share perspective on it thank

you very much for attending appreciate

your time today and have a good one

thank you

[Music]