50. Hybrid and Multicloud Applications are a Headache – Tech Field Day Podcast

Cloud Field Day is all about hybrid, multi-cloud, getting your applications anywhere you want, joining those clouds together and making some cohesive whole. This seems like it's gonna hurt my head, I hope. I really hope that Mike and Barton and Mitch can solve this for me on this episode of the Tech Field Day podcast.

Welcome to the Tech Field Day podcast, where we bring together a group of IT technical experts to discuss a single idea about a key concept within the IT industry. This podcast features a variety of voices from within the Tech Field Day delegate community. And this episode was recorded just at the conclusion of Cloud Field Day 22.

Uh, tech Field Day itself is a part of the Futurum group, and this podcast is also published on our sister company techron tv. On this episode, as we recover slightly from Cloud Field Day, we're discussing the idea that maybe hybrid multi-cloud applications make my head hurt before the discussion. Let's meet who's on the panel today, joining me a complete panel of first time tech field day delegates.

Oh my gosh, that's true. It is. I was really pleased to actually have you guys join us this time.

So We didn't, we didn't know any better. Can I start, Mike, can you introduce just who are you? What do you do?

Hi, I'm, I'm Mike Graff. I'm an in infrastructure architect, uh, based here in the Bay Area. I work for Dolby Laboratories and my area focuses on cloud governance and cloud operations.

Alright. I'm Barton George. I've been in the tech industry for, I hate to say it a little over 30 years.

Um, was at Sun for 13, Dell for 15, and, uh, focused on open source Linux developers. Did a line of Linux laptops and uh, that's sort of, that's sort of where I sit. I'm Mitch Lewis.

Uh, I'm actually part of the FUTURUM group as well. Um, but I'm situated within our Signal 65 labs, uh, which is our, uh, performance and testing group. Um, and I cover kind of all things data, ai, and uh, of course cloud.

One of the, the things that I've seen as a changing cloud, and one of the things I brought to cloud field is the idea that hybrid cloud and multi-cloud are kind of central to how enterprises are really using the cloud. It's not that everybody or anybody is all in on one cloud. You find that people think they're all in on one cloud.

And then there's this exception, I was working at a university that were all on Azure apart from this thing that ran over here on, on AWS and all these pieces of software as a service that are in other places. And my God, the number of external applications they had connected into their active directory that were actually cloud applications, things running all over the world. This really did start to give me a headache, but is that the reality?

Is that actually hybrid multi-cloud and Micah, you seeing hybrid multi-cloud and in the places your work and the places that you talk to? I mean, I think hybrid cloud absolutely right. I think all of us come from a on-premises world and where most, unless you're starting in a place that's completely born in the cloud, you're going to have some on-premises infrastructure.

So I think that is a, it is definitely a real thing. Multi-cloud side. I think it's more what you're talking about, which is I have certain applications running in one cloud and certain applications running in another.

I don't run into a lot of people who are really doing what the, the vendors want us to be doing, which is I'm running the same application across all the clouds. Right? I think there are some very large companies, very large enterprises that are doing that.

But I think my experience, that's not really how it's happening. It's more like, I have this use case that I wanna run on AWS and I have this thing I'm gonna run over on Oracle Cloud because that's, it's cheaper over there for certain use cases. I think that's what I've been seeing.

Are you seeing though, that the, I mean clearly an Oracle database is more likely to, to be the thing I'm choosing to run on OCI as the foundation. Do you see that the entire application then runs on the Oracle cloud? Or are parts of that application running on other clouds and maybe parts of the same application on premises?

I, I think I see more the latter. Yeah, it's definitely like, there, there are certain things that they're gonna be running in one cloud 'cause it's either there has a feature that they want or it's more cost efficient to run it there, but then there might be other pieces or components of that application or microservice that they wanna run somewhere else because it just, it's more effective there. I think a little bit to your point though, your original point is that, um, often, you know, vendors paint this rosy picture of, you know, you're doing hybrid multi-cloud because it's the best way to do it and it's great and when in reality maybe people are doing hybrid cloud or multi-cloud, but it's more out of, um, almost accent or necessity of, well we did this in Azure because that was just the mm-hmm.

Best place to do it, but this is still, we have to do this on prem and this ended up in, you know, AWS or whatever. Uh, so it's more of a challenge than it is the actual strategy. I think it's that that str um, tension you always have between it needing standardization to make things, uh, rock solid.

And you've got people, the developers who are looking for the best tool for the best job, whether he has one. I've got this one, Mitch has yet another. Um, I think that that's what you're seeing in the cloud is people are, and um, Mike said it as you're optimizing for oak, this is great for Azure, so I want that one.

And then, or you may not even know the other person put it together. 'cause your group is thinking of that. And so it's, it's, yeah.

Adele used to say multi-cloud by default, right? It's, it's not by design. You just Sort of happen to be this way.

And I think the not by design element of multi-cloud, I think is there because although Mike, you there are vendors who will say, yes, we'll run, run across everything, but every single one of those cloud vendors will say, run everything on us. 'cause we're the best place to run everything. Well, In an ideal world, that's what you would like to do, right?

'cause it's hard enough to be an expert on one cloud, right? And so if I, if I'm running in multiple, then I've gotta be an expert or have multiples of experts who are focused on each one of those, right? So from a simplicity perspective, you'd much rather be running on one if you can achieve all your, you know, reliability outcomes on that cloud Reliability, performance cost.

Yeah. Yeah, Yeah. Now What I do see is as organizations get larger, the probability that they'll use actively and proactively choose to use multiple clouds gets, gets higher because the number of use cases.

But one of the things that's been a, a message over the last 10 years alongside cloud is digital transformation. And that digital transformation message tends to be all of your applications need to talk together to one another. And that, that's where I see large enterprise organizations getting into this headache of I need to integrate maybe my CRM application as Salesforce.

And so I need to that to integrate into my Oracle database that's running on OCI, but I'm running all of my authentication through active, or I'm sorry, I'm old school, uh, through intra ID. And, and now I actually do have that complex multi-cloud. And of course nothing ev well, we never completely evacuate on premises 'cause that's where our staff are and we've gotta have some it close to our staff and we Still have mainframes.

Absolutely. Maybe we should just go back to mainframes and, and get away from all this, have a single place, a single standardized place for all of our applications. Mm-hmm.

Yeah. But I think it's the, the, you're right in the sense that it's the headache comes from the interconnecting the clouds, right? They don't make it easy.

They don't make it cheap. Right. To do that kind of communication.

And so I think that's where a lot of the headaches come in is, okay, I've got application with microservice a running over here and maybe I've got something over here in another cloud. And then the most kind of efficient, cost effective way for those two pieces to talk to each other is, is the headache. Yeah.

Well, and there's some, some, um, one of the interesting things that came out of Infoblox, uh, the Infoblox team of our first presenter said that they got together with the, uh, DNS specialist from Azure and AWS and the two of them talked completely different language. Yep. They describe exactly the same features with completely different words.

And yeah, it's, it's weird that we as humans have, have done that. Although this is a classic thing for, for a WSI always said, you need to learn to use the words that AWS uses for its things. 'cause they, they don't mean the things you normally do.

And I'm terribly sorry, Mitch, I cut you off in that one. No, Uh, it's fine. That's a good point.

Um, I think another big piece of this headache as you, you know, start spreading out is the security piece too of um, you know, if you're in more places, that's more vectors of attack. That's more things to, uh, monitor and to secure. And I think that was another kind of big theme that we saw throughout pretty much all the, um, presenters.

You know, even the ones that weren't necessarily a security solution or a protection solution. Uh, there's certainly an element of, uh, you know, monitoring availability or, you know, all sorts of those things. What is, you know, what is in my multi-cloud, right?

We saw the, so starting with Infoblox, the story of trying to consolidate to a single tool that will manage across clouds and um, getting into selector of the, having some intelligence that is apply applicable or not artificial intelligence of one form or another, and check out the round table. We're recorded for some discussion about artificial intelligence, uh, but More than one, but Yeah, multiple intelligence as well. I have plenty of intelligences here with me.

Um, we saw that that story around using more tools to manage more complex environments and Catchpoint was the continuation of that. Recognizing there's all that complexity that a heck of a lot of the stuff is carried over the internet. So that was definitely was a, a big theme for us was that complexity of gluing all these networks together and operating and taking away the cognitive load for the operations team.

Also not knowing what you had, I know Haku talked about data all over the place. I don't remember what the number it was, 700. I know some staggering number that they talked to a customer about.

Said, Hey, came back to 'em and said, do you realize you have data in this many places? And they had no idea. So I think same thing is if you're IT and you're looking at, well, where are my apps running?

You're, you would be quite surprised if you went through and inventoried them and had no idea that it was running here. There, I mean, particularly with shadow it, that's the great thing about the cloud. You just put it on your, your corporate credit card and you just, you spin it up without anyone having to, to know and, And going back to security, that becomes a big issue again too with something like ransomware.

Um, okay, you've been attacked, you have data all over, um, you need to get back up and running. What is actually, what is where and what is actually important for me to get back up and running when I'm in, you know, four clouds and I have, you know, all these things that they're talking about. And I think I've got a fresh headache for Mike, because what Barton was describing of the shadow it right, of, of somebody in the, in the team, somebody in in a business team trying to achieve something using their credit card to solve that problem in a way that's not being managed.

Both from, from my perspective, this is unmanaged infrastructure, but one of the things Mike focuses on is finops is making sure that we're not wasting that money. 'cause the wonderful thing about cloud is you only pay for what you use and the terrible thing is you pay for everything you use. Yeah.

Say you just describe my nightmare situation, which is somebody going out signing up for a cloud account with a, a credit card, right? Because then there's no, there's no visibility, there's no governance, there's no Security, patrol Security, none of none of it. Right.

So that's definitely my nightmare scenario that I I I fight against all the time is try, is understanding what's there. And I think that was that I think at least three of the vendors use the term needle in the haystack or needle in a pile of needles. Needles, right?

That's like, that's definitely one of the headaches of a multi-cloud envir or even a hybrid cloud environment is just understanding where things are and and where the needles are. Yeah. I think it goes back to that stability and reliability tension, uh, against agility, right?

So some developer says, I don't wanna wait for it to spin up servers for me. I'm just gonna go and I'm gonna just put on my card and I'll be up, uh, well in five minutes. Um, and so at, at a certain point you're, you're at at odds in the same thing as also there's people who, who believe that yes, we should standardize on everything, except for my case.

It's a little bit different and I think I, I need a separate tool. You guys all use that because that's important to the company, but I need something different. You don't quite get it, do you, That unique use case where, where all the rules don't apply to me, but you better follow the rules.

Exactly. That sounds like executive decisions or possibly the security team. Well, I think that's as a, as a, in an IT shop, if you're, you know, supporting cloud for your company, that's where removing the friction to getting them up and running is how do you kind of avoid that credit card problem, right?

Is if you're, if you're making it hard for them to get going versus then they're gonna go that way, Barry. But if you're removing the friction to getting into the cloud and deploying in the cloud and making it easier for them, then you can kind of forestall some of that. I feel Like because you become more nimble as you, as you get this, uh, cloud, uh, you become in control of that as opposed to having them.

Yeah. And so that, I mean, it forces you to up your game and it forces them to, uh, to start realizing, okay, so there is something maybe a little bit, i, I give a little bit of trade off, but it's pretty good enough for me. Yeah.

And there's, there's some interesting concepts that are really alien and enterprise IT that DevOps practices bring along As a conventional IT person, if I went to the IT manager and said, you know, we are going to take an expectation that there is going to be at some random time, we're gonna be perfectly happy if there's 30 minutes of outage over the next month, right? And DevOps, that's, that's called a failure budget, right? And I can take high risk actions if I haven't had that much outage in the last month, but if I've already had some outages, I've gotta do low risk because that's the level of availability.

Um, and it matches kind of the, the shift we see from on premises IT where the infrastructure has to be perfect, right? You build your VMware cluster so your VMs never go down, um, to cloud infrastructure where yeah, virtual machines disposable. We'll throw that away.

In fact, we love disposable resources. Containers, poof, ephemeral Functions even less. There you go, femoral if there is such a thing.

But that's bringing more headaches for me though, because all of these ephemeral things as an operations guides, where are the logs for this? How do I know what was going on? What was the state of this application last Tuesday when we had a sea of one outage for one of our customers because of a misconfigurations somewhere that affected nobody else?

How can I identify what was going on at that time? And That's, oh, that's your problem. I just need to be nimble and go the way I need to go.

So don't you fix that for me, Mitch, solve this all for me. Don't, please don't tell me I have to run everything on a mainframe and need to relearn for trans 72. You know, I think if I, if I had all the solutions, I'd be a very rich man.

Um, but I don't know. And, and I told you before we started recording this, I was gonna work AI back into the conversation, but it would be A crime not to. Right.

Um, it's 2025. We have talked about a Ai we'll get no funding if we don't. That's Right.

Exactly. But I mean, I do think there's, there's, you know, potentially opportunity there to, you know, leverage AI to help with some of these headaches, right? Um, a lot of it is just managing the sprawl and, you know, logs or, or whatever that are really hard to go through manually.

If you can automate some of those things, um, add something intelligent, you know, to take some of the software plates, um, you know, I I do think that's, um, something that can kind of, you know, act as a, a Tylenol for a headache. Yeah, I would like to prevent the headache by having these tools that allow me to do the analysis without tearing what little remains of my hair out. And it was, was the sort of core premise for selector ai.

It was to be able to use some, um, old fashioned machine learning to predict the state in the future, and, uh, to give you a user interface that's much more friendly by using a large language model. So I think that, that, that concept would, would definitely large language model is a user interface in front of some domain specific knowledge, be it network logs, be it application logs, hopefully a unified collection of network and application logs. So I can get a holistic view and say, why are my queries taking so long?

Or better yet, why are people in? I know, uh, Spain not ordering this particular, uh, towel that I sell because of course, and coming back to it, it's a business question in the end, it's, we're trying to achieve business outcomes with this. Well, I think that's, if you look at both what FortiGate was doing and what, um, Catchpoint was doing, that's kind of the next evolution of their products, right?

Is, is that they're, they're taking those, I don't know what they call FortiGate, called them like they're, you know, low level signals, but they're kind of combining them into something that's adding up to an actionable thing. The next step is being able to ask questions of the, of those signals rather than just having to have them be bubbled up. So I do think that's, and I think, um, Catchpoint is gonna be there as well, is like, why is my application slow?

Right? Yeah. Yeah.

Why is somebody in Turkey complaining about my application being slow? Yeah. Beyond that too, you know, what should I do about it?

Right? 'cause and they're, they're pretty clear that, you know, they'll get you so far, but they're not the necessarily the remediation step. But if you add on, you know, an intelligence layer that can, um, you know, help you work through that, uh, you know, hopefully you have some people in the room that have a, have a good plan of action.

But I think having that extra layer, um, can be helpful. So I think, uh, this AI trend of adding, you know, something you could talk to or something intelligent is something that we'll see across, you know, potentially all of the presenters. Um, I think there's a probably a way to, to work it in Until we have our future where we simply asking questions of, uh, an AI and getting, gaining insights and, uh, we'll get there on our flying cars and mm-hmm.

Uh, all of the other visions of the future. And of course the future will be completely different. All predictions about the future are pretty dicey.

Although in the Jetsons, they did have flying cars, which we don't have, but they did have that giant Google thing that you could talk to and it would answer any question that you wanted. So that was something that did come true. Did it have hallucinations?

No, I'm just talking about just even general Google, but now it's gotten even to a whole nother level. Yes. And then we do get into hallucinations, but hey, would you have put your credit card into the internet in the beginning?

No, you'd be nuts. But now how many times a day do we sign up for things and put credit cards in? And so I think we'll have our credit card moment with, uh, general ai, um, or generative, I should say.

Well, general generative AI soon, And will the headaches of multi-cloud go away or we just change them for new headaches? We're, instead of being concerned about that, the minutia of how our application works, we're gonna be concerned about the minutia of how we construct a prompt to ask the right question. Because it is all about how you ask the question.

Because what you need is a level of abstraction to, to orchestrate this. The thing is, who does it behoove to to do this, right? So zero doesn't necessarily wanna make sure they work with everyone else.

Um, and is there a third party who can do it? And there are people who are trying, but I don't think there's anything yet where you can just push buttons here and there and, and move, uh, move apps or workloads, uh, as we like to see. We're definitely still afraid to let the machine take the action for us in this area.

Mm-hmm. Yeah, I think there's, there's always still more headaches. I mean, you know, you, you asked me how to solve something, I said, well, you know, eventually maybe AI can help.

And immediately we got to hallucinations. There you go. There's another headache right there.

So I think, uh, there's always something else that, you know, another issue that pops up and then there's another tool to fix it, and that's one more thing to manage. So it's kind of a never ending game of whack-a-mole. Well, thank you for joining me for this conversation.

If people who are watching would like to carry on the conversation with you, are there places that we can get in touch with you, Mike? Is there a place, do you hang out on LinkedIn or on, uh, one of the social medias we talk to? Yeah, I Mean, LinkedIn is probably the best way to get in, get connected with me.

com if you want to check that out. I write about cloud operations and Cloud finops and stuff of that nature. But LinkedIn's probably the best way to get ahold of me.

Yeah, I would say used to be Twitter, but I, I've stopped that and now it's, uh, LinkedIn. I'm, I'm finding it growing and really interest, uh, it's great to see a lot of software developers there. That wouldn't have been the, the case five years ago, but that's probably the, the best place to get me is on, on LinkedIn.

I do post a lot to YouTube, Barton 8 0 8, but that's less, uh, interactive. You can, you can put comments on the, the videos, but it's less, uh, discussion. Whereas you can, you can holler at me on, uh, LinkedIn.

Yeah, I'm also on LinkedIn. I have a pretty inactive Twitter account. Um, but then, uh, you can also find me sometimes I publish stuff on, uh, both the future and group website as well as, uh, signal 65 website.

What about Blue Sky? Nope. Yeah, I started there.

You'll Find me on Blue Sky too, Maybe soon. Cloudy advice on Blue Sky, Cloudy advice for all. And of course, I'm Alistair Cook.

I'm the event lead for Cloud Field Day at Tick Field Day, part of the RUM group. And you can find me, well, occasionally you'll find me on the Futureum Group, uh, blog as well 'cause of, uh, the site. Uh, because like Mitch, I occasionally write for that site.

You'll find me also on various places on the tech field. They would website. But if you want to interact with me directly, LinkedIn works, uh, x Twitter occasionally.

And uh, I think those are probably the best places to catch me. I'm experimenting with Blue Sky as well. Well, thank you for joining us for this episode of the Tech Field Day podcast, uh, for more episodes to find all of the amazing episodes that we've recorded and we'll continue to record it seems in perpetuity.

com/podcast, or search for Tech Fields a in your favorite podcast application. You can of course also find this and all of the tech fields, a content across on Textron TV as part of the FU group. I'm Alistair Cook.

Tune in for a new episode of The Tech Field, a podcast next week.