Techstrong Gang – July 16, 2024

Hello everybody, and welcome to the Techstrong Gang. I'm your host, Mike Zern. Alan Shimmel is, of course, still on vacation.

I'll be back next week. But right now we're gonna talk about, well, the state of observability. Then we're gonna dive into what's going on with data in the cloud, because it seems like there's this whole new architecture with our friends from Oracle may be taking the lead on that.

And finally, we're gonna talk about greenwashing with Bonnie. That's right. We'll be back in a minute.

Welcome back everybody. And we're gonna initially start talking about observability, but let me introduce our panel for today. We have, as always, Mitch Ashley over in Denver.

Mitch, how you doing? Very good. Very good.

Happy to be here. Alright. And joining us once again is Guy Courier from the great state of Texas.

Guy welcome. Good to be here. And finally, we have Bonnie Schneider, our resident Green IT expert.

Hi Mike. Good to be here. All right, everybody.

We're gonna start talking a little bit about observability. 'cause this trend has been going on for a while, but frankly, I've been scratching my head about it lately, and it kind of goes like this. So, theoretically, we have monitoring platforms that have been tracking predefined metrics for as long as anybody can remember.

And we were supposed to be embracing these observability platforms that were enabling us to pull all this data that we were gonna instrument in the IT environments, theoretically, using things like open telemetry, and then all that data was gonna be in a place where we could query it and maybe that would surface the root cause of our issues. And yet, a funny thing happened along the way. We somehow rather, are not really driving this at the level of volume that anybody thought.

And in a lot of ways, observability is core to DevOps. But Mitch, what's your sense of what's going on here? Because, um, there's more observability platforms than I can shake a stick at these days, and yet everybody I talk to is struggling.

And there are some big ones too, right? Uh, Splunk being acquired by Cisco, obviously being one of the more recent ones. Actually the active, uh, stack State was acquired by SUSE not too long ago, a couple months ago.

You know, observability is one of those, you know, you can take the same approach and get, do different outcomes. You can throw all the data in one room and say, because it's all in one place, we can go get it and do analysis on it and figure out things that we can't 'cause we don't have access to the data. Well, getting access to it, that might solve it by kind of putting it all in one location and all one tool.

The problem is, is that you need context to be able to understand, to connect the dots, right? Okay. This happened over here.

Is that related to what we're seeing in this part of the software? Which is, is that related to what's happening in the infrastructure? Is that a false positive and nothing to do with it's false flag?

So the, the pro the challenge with observability is intelligently adding context to it so that you can understand what's happening at a, at a macro level and not waste your time looking at things that aren't related, but help you narrow it down to the things of what's really happening. And to do that, yes, you can use, um, obviously Open Telemetry helps a great deal in giving you a common format to get that data into and not worrying about proprietary agents that you still use those or can use those. Um, but you've gotta take all those log the logs, alerts, tracing and that context added to it to understand what you really are dealing with.

And so a lot of organizations take observability and they write kind of applications around it. They're kind of the right apps, right? What's this, what's happening here?

And I think that's where some of the promise of, or the interest in ai, particularly generative ai, to make that easier to query, uh, to get access to that data. Or maybe even write some of those, for example, Splunk, Splunk programming language programs to get you access to that information. Now, other organizations are actually investing in using graph technology of creating connected or organizational graphs or some in some form to add that extra context to, um, what's inside of an observability system.

All that to say, your your average day everyday run of the mill operator network engineers not gonna be spending all their trying time trying to make the observability system work. That's usually what platform engineering and, and, uh, SecOps and folks like that are, are working on. So it is, no, I'm just saying it's no small challenge to take this on.

Um, but, and it's more than just getting data in, in one place as you've gotta be able to kind of thread it together in some meaningful way. That's gonna be helpful. Guy, how you doing?

Doing well. How are you? All right.

Let me ask you this. Um, in your experience, are people really gonna go and find the root cause of an issue or are they just trying to solve their immediate problem and when that gets solved, they kind of move on and then the organization kind of finds itself in something of a vicious cycle because the next person is trying to fix a very similar issue because nobody actually went in and fixed the original issue. And it's kind like I'm trying to just get in and get out as opposed to be, you know, best practices.

Yeah, this is a very long yes or no question. The answer to which is yes, Turn teams tend to be more reactive to problems and proactive when it comes to their aspirations, right? So I think your observation and that question is, is really right.

I do think that most teams wish to maintain, um, a continuous development of, um, uh, of not just the features and capabilities of their app and apps, but on top of that, uh, continuous improvement of performance, user experience and all that stuff. But in the end, the forces that drive them, um, have to do with the delivering, you know, increased capabilities and features. Not so much better latency, but so the, what happens in the end is exactly what you described, um, which is, um, observability and application performance monitoring that's not quite at a minimal level, but, or it's better than minimal viable, but nonetheless is reactive patching and preventing of what happened in the past without the kind of proactive assurance of, uh, new potential problems, especially as they scale or move geographically outside of their core area.

That doesn't tend to happen as much. Mitch, what can we expect from AI in this area? And then one of the articles we have on the site says that generic AI models fall a little short in terms of root cause analysis.

But, um, part of my opening statement was about that I didn't know what questions to ask in the first place. Do I need to know what questions to ask? If I can have AI models that are gonna ask the questions for me?

Is that the next evolution of this thing? Well, we, we were talking about earlier that, um, you know, context is everything and that context is everything in generative ai, right? So the more you can provide either through some predefined parameters in your, in your request into the LLM, defining some context of what you're looking for and why, and giving an explanation of what it is that you're, you're trying to find out, um, it it not only is it retrieval mechanism, if you will, I think that's, that's sort of the not the right way to look at gene of ai.

It will go out and look for the information that you're asking it to find, but it's gonna come back to it in not only entered in a natural language interface, but it's gonna come back to you in kind of people readable information saying, here's the following things that we think are related to what you're talking about. Now, there's some, some belief or or view that generative AI can create novel results. Meaning it's a human that hadn't thought of this one before.

It's not just regurgitating the information that it can provide. The problem is generative AI doesn't know what difference between a novel result and any other kind of result. It's just using, its its own transformer model pattern and matching to be able to come up with a response.

One of the reasons why its responses are non-deterministic. So how will it help us? I don't think it's gonna give us the silver bullet magic answer to say, oh, here's the root cause.

It's going to correlate things for you and give you some kind of hunting tools to help narrow down and define the task. But they may not always be the ones you need to pursue. You're gonna have to so look at it and say, that's false flag.

Don't, don't worry about that. I want to go look at this next. I think that it is also true when it comes to sustainability because a lot of the observability, um, is checking out which applications are demanding the most bandwidth when it comes to energy.

Um, where where can we have more efficient coding? So, um, I think that AI will also be used to, uh, to ask those questions and to get better answers, and that overall will be part of the way it teams and organizations work to reduce energy consumption. Yeah, I just, I don't, I've been wondering lately, particularly when it comes to generative ai, if we need a good metaphor of how to use it, that along the lines of a very enthusiastic, highly energetic, highly responsive, new member of your team who also knows nothing and can do lightning quick searches.

Because my view from the beginning almost has been that the advantage of generative AI is to reduce friction, get you over humps and give you the opportunity to improve quality. But relying on it for any of those things is a mistake. And this is a really good example of it, which is, like you said, Mitch, uh, an AI can think of things or look through a problem more comprehensively, um, than a human can so that when you're stuck staring at a screen for whatever reason, it gets you over that hump reduces friction.

But once you're there, um, you have to scrutinize everything as a human, um, with your experience, with your understanding, with your knowledge. And it may get you to a better place faster, but it simply does not take the place of you or even the most junior new human member of your team. Let's follow that to its nth degree because this is kind of one of the things I wanted to poke at today.

Let's say I have this observability platform today. If I look around it, there's all these silos. I got a security team over here and I've got a IT operations team and a DevOps team.

We've joked about this in the past. There's an ops team for frigging everything. Are we, as we move down the path of observability and we start to maybe create these common data lakes, can we maybe start getting rid of these silos?

Can we flatten this IT organization guy? What do you think? Well, I would say the opposite.

I would say it allows you to take advantage of compartmentalization, to put the euphemistic term on silos that allows you to, um, focus and, uh, identify areas and that need, you know, support or development and focus on them and, and help develop them. And what AI can do exceptionally well is cut across, uh, all of those now looking at those compartments as silos cut across all of them for analysis, for for initial analysis, for recommendations for finding patterns. You just have to know how to use it as, as always, Mitch, you agree or disagree?

Yeah, I, I think Guy brings up a really good point. Um, and we kinda have this situation today where there tends to be two, the largest two camps in observability. The folks that are using it for, let's call it IT ops, the folks that are using it for all the applications and infrastructure, et cetera.

And we also have a very large contingent that are using it for security, um, which is a wide Cisco bot Splunk's. Why they positioned it in the security product and other observability applications or tools, some of 'em not all, um, have done well selling in security space. The problem with that is when you need to bridge those contexts and, and look at the data across more than one domain of where you're using observability, if ultimately isn't in, isn't in the same kind of source of truth that you could then put different use cases or personas or how you're gonna access that information, then you're still limited to what you're doing inside that silo.

So silo allows you kind of customization, but it does limit the data you might have access to. Uh, and you can take approach. I'm not saying it's easy to get to one place where the data is stored, even in a world of observability.

If you can bridge those two together and kind of not make the data the silo make how you operate can be specialized, if you still wanna call that a silo, then, then you can get kind of the best of both worlds. The guy's talking about. To, to touch on Bonnie's earlier point, there are so-called silos or compartments that occur culturally or socially or, or organizationally that you can open up to this sort of an analytic level.

Um, and I'm thinking of sustainability in particular, you know, GRC in general governance, uh, uh, risk management and compliance that these teams don't even want to think about. Probably never think about. I'm not sure I entirely agree.

I can see a world that goes something more like this, especially in organizations that can't afford specialists. There's, uh, economics that factor here into the single biggest cost of it is still all the people involved. Can you not imagine a scenario where it says, I have a core set of human IT operations teams, but all the specialists are really AI agents that I'm just invoking to handle a specific task that I previously had to have a really expensive specialist in the form of a human to go do.

So I can see a world where, um, the AI agents are part of the IT operations workflow, but um, they're the specialists. So I know that we said that the AI agents are stupid right now, but they won't be stupid forever. I think that's especially true with smaller companies and smaller businesses maybe that don't have the resources like you're talking about Guy.

Well, I think of the, um, the thing that's cropped up recently, which is, you know, I don't want AI to come and do my main job, but that I can do the dishes. I want AI to do the dishes so I can go back to doing my main job. I think that there's a real, I would call it a danger in what you are suggesting Mike, um, of, uh, devaluing or misunderstanding the role of general inte intelligence that the human, the human factor has.

Um, I don't see how, uh, an AI that specializes in an area, um, where the IT generalist is not specializing absolves the IT generalists of un understanding, learning and becoming better in that specific area. As a question of supervision, now we're talking about ai, not a gi artificial generalized intelligence, which is five to 10, maybe 15 years away. That might be the difference there.

But for now, literally what we're talking about when it comes to generative AI is not artificial intelligence, but the simulation of intelligence by artificial means. It's a simulation. We'll see.

Hey Mitch. Um, one of the caveats of observability in my mind is that, um, I have to have agent software in the form of open telemetry or something coval and thereof that I'm gonna, uh, include in almost every application, maybe every piece of infrastructure. Um, to me that suggests that, uh, it's complex.

And a lot of folks I've talked to say, you know, the agent software is as difficult to deploy and manage as any other piece of software. So it just exponentially creates additional load. Can we instrument things better?

And are there other ways of thinking about this? Well, I think where the complexity comes in, Mike, is you have, if you have to, um, think of, think of it this way, if you have to install an an open telemetry agent, if we wanna call it that, along with a proprietary agent. Now what Open Telemetry really is, is a way to communicate in the standard format the data that's being gathered by any kind of an agent or a piece of software.

I mean, it couldn't be your application that's emitting information, uh, that could be logged into through Open Telemetry and then go into an observability system or other analysis tool. So it's, its purpose is to kind of decouple the proprietary agent from the data that you're gathering. So vendors are not competing on, you know, the, uh, best feature of their agent.

'cause compared to everybody else, I think at practice what happens is people oftentimes end up running the agent true 'cause, uh, two 'cause it may have some control properties and otherwise, um, I don't think the complexity comes from, um, the Open Telemetry agent necessarily until you, unless you get more into some of the architecture of Open Telemetry. Um, it's a really good book by Austin Parker, um, on Learning Open Telemetry. I would recommend, I think it's an O'Reilly book where does a really good job of kind of explaining how open telemetry works, both not just from a developer or someone writing an agent perspective, but how you operational operationalize it.

Um, it might help with some of that complexity, But Mitch, don't you think Oel helped alleviate a significant burden of observ observability platforms, which is that they have their own infrastructure needs, usually several stacks and kind of a, they have an application life cycle of their own. And by removing, um, the, uh, the need to manage the, let's call it the integrations in general terms or not all of the need to manage them. I mean, when you can go agentless, I always perk up my years and, uh, hotel allows, uh, basically, you know, the application vendors to pre-build or for you as you're building your application to pre-build it for an observe any observative observability platform, and that takes away that infrastructure operations spurt.

Yeah, I think, I think your point's a good one guy of in a, in a proprietary stack at some point that what that integration means, part of that integration's transformation of data, right? How you get that data out of how that vendor represents it into something that another tool can understand, whether it's the one-to-one integration or a one to many. And that's what Open Telemetry is kind of a one-to-one, one to many where the vendor can still log data in their own format, but they can emit that data to an o to an observability system in, in an hotel format or an Ortel hotel mechanism.

Uh, then can be that data then can be used without, you know, a lot of further transformations by other tools. So you, you can still have a proprietary stack, um, even including agent that's collecting data, uh, that's then later emitting it into Tel. Or you can kind of have that bottom of the stack, that agent part of it, really just be go tell native and that way you don't have the issues around at least that part of it, the proprietary stack in the agent side of it.

So it can, it's a mixed bag. Depends on where people, when people come to hotel, if you're coming to it later, you probably not gonna rewrite everything on day one and hotel. You're gonna use what you got.

Right. And All right folks, I don't think we're gonna resolve this issue today, but I would just point out to you that the combination of AI and observability is gonna have a profound impact on it. We're just all not quite sure exactly how it's gonna turn out, but we'll be back in a minute to discuss the next amazing advances in technology.

Cloud native now is the web's leading resource for the growing cloud native ecosystem. com is your destination for news, thought leadership, features and webinars on cloud native architecture, Kubernetes, serverless, cloud native application development, microservices, service mesh, cloud native security, and more stay on the cutting edge of modern application development at cloud native now. Alright folks, and we're back in, well, we're talking about data and data storage and I gotta say, I cut my teeth on this space about some 25 years ago.

So it's a subject that's always been near to my heart. Oracle has a new platform out called, um, Exodata Scale or something. You're equivalent thereof.

I'm never sure of their branding names 'cause well, they're hard to follow. But the basic idea here is that you'll have a pool of storage that sits underneath the database, and this is gonna be highly dynamic storage. It will include memory, it will include flash drives, it will include disc drives, and they all look like one common pool that you can access and in bulk.

And the platform will automatically move data to the various, uh, layers that make the most economic sense based on the access and what's required. We've been talking about this conceptually for a long time, but, um, guy, let's start with you. What's your sense of what's going on here and, and I feel like, didn't we get here already, but now we're doing it in the cloud?

What's your sense of our maturity here? Oracle definitely has been here for years and years, not in the cloud. Um, they have been selling, um, what I think Gartner called, uh, integrated stacks, no, IDC called integrated stacks, uh, for 10, 15 years, something like that.

Um, the names have changed and of course, uh, Oracle has been busily, um, very busily helping transition its entire customer base to cloud-based delivery to the degree that their customers want to, but they have never given up, given up this full stack approach. And so the, the, the, a new service or the new capability called exascale, it is for exit data customers only. So you have to be an exit data customer to begin with.

Exit data is the global, um, the multi tier, multi uh, uh, mul multi, um, technology, uh, storage environment for Oracle Cloud infrastructure customers. What Exascale does is solve a really important problem for Oracle customers and really for enterprise customers generally, which is how to get the right data and metadata to the application, um, wherever the application is being run or used as rapidly as possible. Um, as our friend Keith Townsend likes to say, um, the fastest data migration is the data migration you never have to do.

So if you are running an application either on Exadata or typically an Oracle application is now are often running on exit data, um, all the way down through the Oracle database, down to the Oracle cloud infrastructure. Um, you have this data distribution problem, you don't want users to have to wait, uh, at some point for some data become available. This is particularly acute for AI applications as everybody knows where context is, once again, king and providing the greatest amount of content provides the best overall results outta the AI and experience outta the ai.

And so what Exascale is doing is it is splitting out all the storage in all its various forms, like you say, from spinning disc, uh, up to, you know, really in memory, um, uh, providing a highly duplicated globally available metadata structure, very high availability, it's reduplicated. And then adding, uh, putting this on, uh, the Oracle 23 AI database for AI purposes. Um, adding, predictive, querying, all these things just to increase the speed, um, of and, and the reliability of services that rely on it.

And I will add just a couple other things. Um, so to go with the fact that you need to be an exudated customer to take advantage, um, you're actually now with a much simpler, uh, service set. Um, you don't have to worry about ingress or egress or networking charges or any of that other sort of stuff because exascale handles all of it for you.

That's a radical simplification of most enterprise data and enterprise database application, um, uh, you know, management. It's really helpful. All right, Mitch, here's the part that makes me scratch my head.

You know, I'm, if you noticed, I'm doing a lot of that lately. I'm, when I was younger, we always said that you bring the compute to the data and far frankly, nothing good could ever happen when you move data. Um, and then the cloud came along and suddenly we were all moving data into the cloud and everybody was going, you know, that was the way to go.

We were gonna get rid of the data center, et cetera, et cetera, et cetera. It almost seems to me in the last year, I keep hearing this phrase move the compute to the data again. So, but the data's still, the bulk of it resides on premise and some of it's in the cloud.

I mean, data has gravity. So your a sense of what the heck is going on. Well, yes, data has gravity, Mike, um, it also requires governance.

Part of the challenge is, while a lot of it is in the data center, maybe the data center, local data center or the data center in the cloud, um, you know, applications don't live, um, on just code alone. They, they both feed and create data. So oftentimes this data is, is not just in on, on your local premise.

It could be created at the itch could be all over. And that's part of the challenge with data is like, where is this? Who's using it?

How's it being generated and should it be where it is? And I think one of the challenges in, in data management is you have all these different mechanisms of, or services for storing data. And, and what the, what the Oracle model of the exabyte model does is kind of bring something you mentioned earlier, very familiarity, or sim being simple as guy described.

It is it looks like a nas like a network attached storage. It looks like a, you know, here's my end memory, here's my solid state storage, here's my, you know, fast retrieval, here's my medium and long-term storage so that you're paying, you know, different amounts of, of dollars depending on what your retrieval needs of that data. Now in the cloud that's distributed, right?

That could be in one, uh, one center operational center could be in multiple within that provider. And same true with with Oracle. So I think here is, you know, there's, from a developer perspective, you always think about don't worry about what the data is.

I'm just gonna use an API or a service to get to it. And then the people that manage the data and have to operate it have to worry about those things like retrieval rates and cost and storage, and where am I keeping that stuff. So hopefully the Oracle Exabyte solution, I think brings some simplicity just for that reason that it looks familiar.

Hubs operates familiar to, uh, data stores. They're used, used to OnPrem. You made the perfect point there too, Mitch, which is, it looks like a nas it looks like, in fact, um, in a typical Oracle model, you are, um, it's just sort of as like a pure API access, so to speak, pure, um, where, um, from even the infrastructure operator standpoint, they no longer, they just looks like one, uh, data store that they're, if not database that they're managing, and they simply do not worry about what happens underneath.

I think that bigger worry might be something Mike alluded to, which is how do you get the data in there to begin with? You gotta buy the exudate service, of course. And are you gonna have to do ETL?

Are you going to, you know, and are, do you need to use Oracle itself? Uh, you do. Um, so how are you gonna manage that ingress of what is probably either, at the very least distributed data and at the very worst, some giant multi terabyte data stores somewhere Oracle has services to address this sort of thing.

But again, you're starting to get more into this oracle world, which may not be your choice at that moment. So I have a second thing about this that's making me scratch my head. And, and Mitch alluded to it.

So we are seeing more processing of data at the edge, right? The idea is that we wanna bring more of the processing and the analytics and apply it to the point where the data is being created and consumed. Makes perfect sense to me.

It's kind of, we want this near term or near real time interaction with the data. But if that's the case, then what is the role of the cloud going forward? And, um, am I just taking the aggregate of the data process at the edge and storing that in the cloud as a big repository, but all the action is gonna be at the point where the data's consumed.

I, I kind of feel like, you know, maybe batch joining applications are going away of the dodo bird. I mean, guy, you want to take a stab at this? Yeah, I think a lot of what you're talking about to me kind boils down to the perpetual tug tug of war between the application developers and the infrastructure folks.

Frankly, it should not matter anymore from the developer standpoint, um, where whether it's edge based processing, it, it should matter in the sense of what you tell the application to do. But the idea is that the availability, the, uh, latency or low latency reliability of the system is left up to the ops folks, the app dev folks code accordingly, right? So what you're describing sounds very old fashioned to me, which is to say, well then don't you need a data store over here at the edge?

Um, don't you need a data store, you know, in this region versus in that region? And if all of Oracle's moves over the last maybe two years, say anything, they are trying to solve that problem. Not to take away all the operations issues, but allow the operations folks to work it, uh, you know, interactively with the app dev folks to understand, um, where the user bases are, where the data is being collected, where it's residing and all that other sort of stuff.

And just sort of twiddle dials in the XA data or in the Oracle, uh, control panel that ensure that the data or metadata are where they need to be at the time the applications run and the rest. That's how the cloud works, Mike. I mean, it's not supposed to work, so it's not really more complicated than that.

Mitch guy just said, I'm old fashioned, which is not untrue. But let me ask you this, Mitch, don't I wind up just putting Cashs everywhere to kind of make up for the latency fact? I mean, the laws of physics have not been suspended even in this newfangled world.

We live in It, that that is still true, right? Replication, caches, things like that. And so, and it goes both directions, right?

If you are doing work at the edge that you might require an LLM or SLM, um, that might need to be fairly local near the edge or at the edge to be able to do the kind of processing that you're looking for. If you're collecting data from a car, the telemetry could be gigabytes of data, terabytes of data that gets collected on a daily, weekly basis that might need to be transformed or analyzed, uh, locally, but also then replicated or transmitted to the, to the larger cloud. I think, I think the point is, which is what point guy was making, is you wanna be able to talk data through a service, whether that's a microservice that I'm writing or it's a Oracle service that I'm calling or whatever that service looks like, as opposed to assu assuming data is there unless you know that you, you have specific processing or latency throughput requirements.

And then, then it's an architectural decision about where does the data need to live, be replicated, et cetera, transformed into a different subset so that I can use it at the edge if that's where I need it, or that's where I'm consuming it. And that way your code, your software developers don't get wrapped up into the, oh, I'm just gonna assume everything's local and I'll just kinda live off of that. Well, you might write it that way, but that service is actually going to talk to something else that they'll transmit the data that you want for that query instead of having the data live locally.

So I think you're being old fashioned is not a bad thing, Mike. We want to use some APIs and some other good things like that to, to talk to data as opposed to making too many assumptions about, 'cause where it is today may not be where it is tomorrow. That's, That's a great point, Mitch.

I think that one of the things I'm, I've been seeing with the companies I've been talking to and interviewing is kind of having that combination looking for efficiency in the cloud for where they're storing data, but also creating innovative new solutions for building new data centers and how they're going to run those data centers in innovative ways. I mean, just recently Microsoft had an, an endeavor where they were looking at underwater data centers and, and so new ideas are coming out with how can we, um, power these with renewable energy, but how can we also find efficiency in the cloud? But I also have a third thing that is scratching my head directly related to that.

Okay, so the third thing that makes me scratch my head is we're talking about sustainability and we're gonna be more efficient, and yet we're gonna be processing more data than ever outside of these cloud environments. I mean, AWS last week was talking about they achieved some sort of, uh, carbon neutral standard that they were pushing for for 2030 this year and golf applause to them. And that's nice.

But, um, more data than ever is gonna be processed not only between at the edge, but everywhere in between. If you listen to what Mitch was just saying, we are gonna be processing data, uh, at the edge of the network and probably as it streams back to the cloud and then the local data center, and then we'll get to the AWS environment. So I just can't help but wonder, are we kind of like kidding ourselves about our sustainability goals?

'cause we're not really taking into account all this data that's gonna be processed somewhere else. That's right. And not to mention the AI that's gonna be used to, to process it.

So I think that that is walking that, that balance that all of these companies as big tech and even smaller companies are trying to balance, uh, we can do a lot more. We also have to use less energy and we can't, um, we have to offset our carbon emissions. And that's not just, you know, leasing that, that's a lot of new regulations that are coming into play, uh, in 20 20, 20 25 and 2026 as well.

So, uh, I think that the balance is there and some companies are finding it through balancing those, um, those emissions, let's say with Nvidia, with having more, um, efficiency in, in, in their, in, in their process. So it depends on the company, but I, what I've been seeing is a combination of factors to get these companies to carbon neutral, carbon zero by certain dates, and it comes to a compilation of different, uh, maneuvers, whether it's a re renewable energy for a data center in more efficient operations in the cloud, or even that circular, um, packaging and things like that that we talked about last week. All right, given all of that, Mitch, I'm coming back to you and we're gonna go with that old fashioned, but back in the day there used to be people called enterprise architects.

Are these the new kids that are cool again, that their stuff is so old, is new again, because in to make any of this work, it seems like you need a defined a architecture. Um, they haven't gone away, we just haven't talked about 'em in a while. So maybe we've been talking too much about cloud architects or platform engineers or other topics, but, you know, and that's, that's the interesting thing about when we talk about generated code and, and AI is yeah, but as be in the context of some of the larger system that people have really engineered and will be engineering for a long time.

That's, that's, that's the role of the architect is kind of make all the pairwise connections of all the functionality, whether they be through apps or third party SaaS services, third party cloud services, internal databases, and recognizing where transactions flow across or data flows across all those connections and trying to optimize that in a, in an architectural pi, uh, paradigm so that you'll get the kind of through throughput la latency efficiency that you need for the application. So if you're suddenly, if you know you're pushing more out to the edge, more processing, it's gonna be demanding more data at the edge. All right, so what do we do with that?

How do we replicate data? How do we cache some of that locally? Um, and how do we, how often do we refresh and keep that cache, um, alive so that it can do what it needs to do at the edge?

Or maybe it's, you know, kind of a time of day if all of the sun type of application where that shifts and we need to migrate data around to support it. So that, that's all the kind of keeping the system context together and understanding how this organism needs to work and operate and breathe. And then working with the developers, working with the operations, the cloud architects, the platform engineers to, uh, help you, you make sure you know what you're doing in terms of releasing new capabilities that the underlying architecture is gonna support that.

All right, guy, you want to add last thought here? Last? Yeah.

Um, I wanna touch on a couple things that, uh, us old fogies said. Um, uh, Mike, you talked about, uh, earlier on you were saying, well, now don't I have to have a cache everywhere now I have to have a cache at the edge, now I have to have a cache in the data center wherever. Right?

And then Mitch, you were just talking about, uh, in the context of enterprise architecture, that one of the things the enterprise architect needs to do is to operationalize, uh, I'm sorry, optimize in an operational paradigm. So One of the things the enterprise architecture does is figure out what the systems are, not just vertically and how they connect, but I mean, it's necessarily not just horizontally and how they connect, but vertically as well and who is managing what, or at least you know, what functions manage, what are we gonna do that, you know, so that you can make decisions about whether a service provider or a vendor or cloud provider or whatever is going to be handling certain things. So one of those things is cache.

It may not be across the enterprise, Mike, but on the X data, the excess scale announcement shows that a vendor can handle that for you. So actually the answer to your question, don't I have to XY, Z with my cache? The answer could be no.

The enterprise architect needs to build that in, needs to explain that. And overall year period, maybe all of the caching, all of the availability concerns get subsumed by the vendor for a particular application or class of applications. You multiply that across, add AI into the mix, of course, and the enterprise architect's job becomes more important than ever, but it doesn't mean that more and more of the operational burden can't be handled outside of the IT shop itself.

So the IT folks are focused on focusing on proverbial higher level things, including when it comes to storage. All right, folks, I think we've got this topic kinda locked down. Essentially we're saying that you're Still scratching your head though.

Oh, obviously. 'cause enterprise architects are kind of like my dance moves. They're so old, they're new again.

But anyway, we're gonna be back in a minute talking about greenwashing. Hi everyone. Welcome back to the Techron gang.

We're talking about greenwashing, and you might be wondering, well, what does that even mean? Have you seen terms like eco-friendly or biodegradable on just products that you buy? Well, it's a little bit more intense than that because it's not just those little labels, but what do they mean?

And are there any repercussions for companies that are not honest about what they're saying? Well, if they say those things and they don't really mean them, and they can't have the, they don't have the evidence to back it up, that's called greenwashing. And it's coming into play for tech companies and for major corporations in a large way, due to new rules that are being released by the uk, particularly their financial conduct authority, that's going to affect tens of thousands of companies.

Those rules are in effect now. And in the this video, I break down what it is and why it's happening. Hi everyone.

I'm Bonnie Schneider with your Ego Tech Analyst insights. Today we're diving into the UK's new groundbreaking anti greenwashing regulations. What is greenwashing?

It's when companies overstate or make false claims about the environmental friendliness of their products or practices. The UK's new anti greenwashing rules apply to any financial conduct authority regulated company promoting products or communicating with UK clients. Regardless of industry.

This is a major step forward in holding companies accountable and safeguarding consumers from being misled. This could range from exaggerating a product's benefits to outright fabrications. No more vague eco-friendly statements without solid proof.

Critics of these rules though, do cite several challenges. For example, additional costs may come with investments in technologies like AI and blockchain to track and validate these sustainability claims. Companies may find it difficult to navigate complex reporting requirements to ensure compliance, but with challenge comes opportunity.

Companies that make proactive efforts can leverage this regulation to shine a spotlight on their genuine commitment to sustainability. Business strategies to meet the moment include investing in notch data systems to ensure accuracy and reliability and boosting credibility. Using a third party source to help monitor a company's claims, eco-conscious effort like bees will play a crucial role in shaping the industry's response to climate change.

Now, you may be wondering, what do tech companies have to worry about this for? Because the financial conduct authority of the UK does not oversee tech companies. But here's the thing.

Banks, um, investment investors, insurance agents, that's where it comes into play. So any tech company that is dealing with those types of financial organizations, and most of them are, are going to have to answer for things like their packaging, their, um, supply chain, their processes, are they greenwashing this? And the changes that if they are, that have to come into play are pretty widespread because that affects the marketing, uh, communications, everything, all of their messaging.

So I think that for this is the beginning of a lot more regulations to come, but the greenwashing rules that are affecting the financial in organizations right now are affecting tech and many other industries right now. So guide, are you seeing companies actually like get on board for this climate change thing outside of Europe, per se? Um, I talked to one fellow in the us we were in Barcelona, and I've asked him about this specifically, and he basically said flat out, no, There's a real sharp divide.

And I would actually divide the divide again. So the sharp divide amongst the companies I talked to is in the maybe upper end of the mid range in size, in revenue, um, and up plus, uh, organizations of all sizes in the public sector. It's a concern.

But the second divide I would put is those who are, uh, taking a, a focus on it for marketing reasons versus those who have some other sort of, um, related agenda. I mean, energy companies in particular, I think, uh, are sort of notorious for, it's not necessarily greenwashing, but, um, seeming to pursue, uh, these initiatives more for marketing purposes, uh, because they're thought of as polluters, um, un unless they're in the clean energy, um, side of things. Um, but I, I find that it's really more of a cultural thing.

Uh, the minority are genuinely, um, outside of their industry, um, as well as, uh, when they're in the public sector, um, not they, they're genuinely looking for good solutions to this sort of thing. There's a major shipping company, for example, that might be the one that make the same customer you and I spoke to, um, that, uh, take it as, uh, really a requirement, uh, for them, um, really for business efficiency purposes, for marketing purposes as well as to, um, be better, you know, uh, world citizens, global citizens. Um, but I think that you're right.

The, the majority of firms are either too small, um, or greenwashing and greenwashing similar activities are still attractive to them because really what they want to do is have a good message to the market. But, uh, I think that that's changing because of these new regulations, the one I just spoke of with the financial conduct authority, that's only been in effect for less than 30 days. And then we have new, um, regulations that will impact tech companies, uh, globally right now, the largest ones are being affected, but coming up next year, which is less than six months away, we are going to see new regulations where not only will will the marketing have to be accountable, where you can't just say something and it's greenwashing, but they're gonna have to do a lot of ESG reporting, reporting on, um, how their carbon emissions, their social governance and things like that.

So I think that where it started from, well, this is good press, we wanna do this, we wanna show we're responsible where it's going to shift in the next six months to, I'd say 18 months for sure. Um, the necessity of it for regulations and accommodating regulations. Some mostly in Europe, but also in California, new regulations are coming up for reporting.

So I think that shift from, it's a nice idea to, well, we really have to pay attention to that is happening now and it's going to shift even further in over the next year. Yeah, I'll be honest. That don't happen.

Go ahead. What's your sense, Bonnie, of the systems in place for this? Because this raises yet another frontier in the compliance efforts and data providence, accountability and, uh, uh, avail, but not availability.

Um, the ability to create attestations for this in order to satisfy the authorities and the new regulations are, are the tools there? Well, that's a great question. Price Waterhouse just did a, uh, big survey with executives from all over the world and I think 30 different countries.

And most of them said, yes, we are ready and we have the tools, we'll be fine. But then as they dug deeper into that survey, they found, well, we don't have all the tools in place and we still have a ways to go. So some of the larger organizations are relying on third, uh, are they have internal systems, like the big tech companies to monitor and be able to regulate it.

But when you start getting into, uh, the smaller to medium size, uh, entities, not so much. You know, that's where, that's where I think there's gonna be a lot of education that happens, um, internally and externally. And perhaps that depends on third parties to, uh, calculate and monitor things like carbon emissions and looking for ways to be more energy efficient.

To your point guy, and yes, again, I'm scratching my head, um, I don't understand how to correlate the amount of energy consumed directly to the amount of carbon generated if I'm running servers somewhere. I mean, is are there, have you seen any tools to that level of detail? I'm not so familiar with the tools, which is why I asked the question.

I think there's a fair amount of proprietary work being done on this. Um, it has to be sort of legally defensible, visible at testable. Um, I know that, um, there are a few open source projects or openly available projects to do just this sort of thing.

And there are certainly, I'm not familiar with the particular regulations or terms there are, uh, globally accepted as well as, you know, in some sovereigns the contexts legally accepted calculations for this sort of thing, when the data is centralized, and particularly when the systems are centralized like a bunch of servers or racks or even just a edge cluster. Um, and you can, uh, with the sort of auditability, you can attest a certain amount of energy usage. Uh, there are formulas that exist, accepted formulas for converting that into carbon use because ultimately the sources assumed to be, oh, well, you actually have to provide the source, whether it's a cons, uh, renewable or non-renewable source.

And then you do the calculation and that's it. That's the state that we're at right now, as far as I know from the customers I worked with. I think that is exactly right.

And I think that explains that survey image where when you dig into the weeds, it's not so easy. You know, it's that process of, of it being much more complex and involved, and then you have the time el element of having to figure that out in, in a short amount of time. I'll give you an example of some greenwashing that goes on.

Somebody stands up and says, uh, we use this amount of renewable energy, and then you peel underneath it and you find out that they bought the renewable energy credits from somebody else and included that in their assessment. And so it's not like we had a net reduction per se, we just kind of moved some financial engineering around. So how much of that is going on?

Probably a, a, a fair amount. And I think that as these, um, these regulations come to play, as I mentioned, and measuring the output, even when it comes to ai, we all know it uses a lot of energy, but we don't know. It's, it's very hard to calculate, let's say.

So I think that that's something that's gonna be emerging as we go. I don't think there's, there, there are some obviously defined formulas for it, but there will be some flexibility, um, because as the technology develops, Is that greenwashing though, Mike, if I'd buy the carbon offset, it's being accomplished somewhere in an audible and traceable fashion. So when you buy your, uh, uh, flight, um, back to New York, Maine, London, wherever you're going on vacation next, um, when you buy your flights, if you choose a carbon offset, there is some sort of data chain or information chain through the airline going back to somebody somewhere planting trees or whatever they do.

I don't even know what they do. So that's not greenwashing because you can prove that somewhere the carbon was offset. Mm-Hmm, makes sense.

The fact that you didn't do it yourself. Yeah, I think that going back to your marketing point, um, what I just described is usually like in the finest of the finest of the finest print at the bottom of the thing that actually says that. So it's not like they're being completely transparent about how this, uh, offset is being accomplished.

Um, I believe if, you know, I don't know if it's probably changed, but a lot of the Tesla financial numbers, you know, that includes carbon offsets that are being sold to other car companies, right? That's right. Yeah.

And I think I can imagine, um, sort of a maturing of the terminology and the understanding. So you could call yourself an originating, you know, uh, carbon neutral as opposed to just carbon neutral or something. I dunno.

It's a fair point. It's a fair point. I think that clarification, um, and then breaking down the, the descriptions as as we go, definitely will happen.

I like that. What do you think, um, can we put that in? Let's get there.

Go. Yeah, I'll call the uk. I'll get it in.

Alright, let's get it going. So what is ultimately your best advice to organizations that are a, trying to position themselves as being green, but, and then b the other side of it is people who are trying to put specs in place that says we're only gonna buy stuff from people who are green. I think you just want to, uh, approach it like you approach any business opportunity or, or, um, scenario you want to have a cost benefit analysis and you wanna systematize it.

The one thing I would say is that your company culture, internal, your brand image, um, how you are working with the world, that's part of your cost benefit analysis. So unfortunately the word cost is strongly associated with money, benefit also with money, but be a little more expansive. And in fact, if our work in cybersecurity has taught us anything, there are soft costs, soft benefits that you can ultimately turn into a a number.

Uh, if you wanna do that, just be pretty expansive there. And if you are secretly wishing as a business leader, um, that your organization is more green, you can make plausible, um, arguments that accrue to the bottom line out of that position and posture. But otherwise it's standard business exercise.

Well, even if you're a nonprofit, Does this always gonna come down to regulations or will people just start doing the right thing one day because well, it's the right thing? Well, I think that that's where it started right now, where a lot of people are doing that, but that regulations are gonna give it that extra push, especially for the larger companies that are doing business in the eu, um, and, and California because those are some regulations that are coming into play and, and throughout, uh, the Textron gang in, in the coming weeks, I'm gonna have more analysis, um, on some of these new regulations that some of them, a lot of folks may be not aware of and we'll break down how they might affect tech companies. All right, folks, I think we are running outta time on this subject and all our other subjects for that matter.

As you can see, I scratch my head a lot today and well, um, that's a regular routine for everybody who knows me. I want to thank you all for watching the latest edition of the Textron Gang. Please stay tuned for all the great episodes we have behind this.

Until then, we'll see you next time.