What Hybrid Infrastructure Really Means with Matthew R Downing

WEBVTT
00:00:18.853 --> 00:00:29.321
Hi, I'm Paul Price with ISA, and this is the Argument my podcast and webcast series with architects from around the world.
00:00:29.321 --> 00:00:40.295
Today, we're going to be talking with Matt Downing about hybrid infrastructure, and I could probably go on for an hour on this podcast.
00:00:40.295 --> 00:00:42.021
But, matt, why don't you introduce yourself?
00:00:42.021 --> 00:00:45.871
I've been lucky enough to know you for a while and you have to tell everybody about the cats.
00:00:47.661 --> 00:00:48.442
Okay, brilliant.
00:00:48.442 --> 00:00:50.970
Thanks, paul Matthew Downing here.
00:00:50.970 --> 00:01:07.066
The 30-plus years in technology starting at university level I was doing chemical engineering Fell in love with technology when I got to do a whole lot of things like MATLAB and various programming languages to solve engineering problems.
00:01:07.066 --> 00:01:24.993
And over time, the techniques I've learned there, plus problem solving techniques I learned in engineering, has really helped me drive a career through IT and technology, which has now ultimately led to me going that architect route.
00:01:25.180 --> 00:01:29.487
I didn't want to be a specialist in one particular area, you know, I didn't want to be a developer.
00:01:29.487 --> 00:01:31.587
I didn't want to be a systems admin.
00:01:31.587 --> 00:01:38.751
I wanted to always play the whole enchilada, if you like, and by doing that I felt like I had more connectivity with people.
00:01:38.751 --> 00:01:51.984
I got to understand the way process and business works, what the purse string holders care about and, more importantly, what the purse string holders care about and, more importantly, what the customers care about and how does technology enable that.
00:01:51.984 --> 00:01:55.893
So that's kind of how my career has developed into architecture.
00:01:56.801 --> 00:02:13.308
What I've seen more recently which has really been amazing, with the whole advent of cloud and the adoption of really hosted systems in different data centers, which is cloud, and then the evolution further into clever data models, clever networking.
00:02:13.308 --> 00:02:16.655
You know AI is coming in clearly strong.
00:02:16.655 --> 00:02:25.174
You know all of those have just really feathered that cap around technology, delivery and business value.
00:02:25.174 --> 00:02:35.324
I think that this hybrid course is probably a good timed course in terms of people have had a taste of cloud Good, the bad, the ugly.
00:02:35.324 --> 00:02:38.967
People have probably spent a fair bit of money on budget.
00:02:38.967 --> 00:02:46.889
People have probably experienced the whole shift of skills you need to do and you know all of these sort of things play into the dynamics of it.
00:02:50.461 --> 00:03:05.353
So I'm going to ask you a question, and this, I think, is because we've spent all of our time on, you know, public cloud versus old data centers, and I just think that that is a very nearsighted question.
00:03:05.353 --> 00:03:11.610
In 10 years, how many locations is a corporation going to be running compute?
00:03:13.382 --> 00:03:19.028
Well, it really depends on their customer model that they are trying to reach.
00:03:19.028 --> 00:03:30.433
So you know, if your market is distributed all over the show and you really want to reach them, and if it's really content intense, you may have quite a massively distributed network.
00:03:30.433 --> 00:03:39.514
If it's not as distributed or if it's very low touch for customer, you might actually still remain in a fairly centralized model.
00:03:39.514 --> 00:03:41.006
And there's everything in between.
00:03:41.006 --> 00:03:43.103
So, between the sublime and the ridiculous, really everything in between.
00:03:43.103 --> 00:03:44.185
So between the sublime and the ridiculous really.
00:03:44.205 --> 00:03:48.831
But this is where I really get into the question of what does compute mean anymore?
00:03:48.831 --> 00:04:13.058
In an edge world, in a world where smart devices, in a world where literally everything has a processor, we're used to thinking like big IT, which just means we think about finance workloads, hr workloads and boring workloads right, but manufacturing has workloads running in devices.
00:04:13.079 --> 00:04:24.882
This big right, and I'm not talking Raspberry Pis, I'm talking people's shoes and belts and cars and their handheld specialized equipment and satellites.
00:04:24.882 --> 00:04:41.500
Right, what does it mean if the world of infrastructure for a corporation needs to expand to infrastructure for humanity, right to roads and buildings and stuff like that?
00:04:41.500 --> 00:04:43.932
You know what does that mean for us?
00:04:43.932 --> 00:04:49.235
How does that look to an infrastructure architect or to a person thinking about this right now?
00:04:49.235 --> 00:04:49.755
I think?
00:04:50.598 --> 00:04:59.146
for the ever optimist, lots of excitement because there's a connected world and you can get to everything and everything's abstracted and you've got all this great stuff.
00:04:59.146 --> 00:05:15.132
For the slightly more skeptic probably a few sleepless nights as you're worrying about costs, you're worrying about operational impact, you're worrying about security and I think you want to find that good, comfortable space in between.
00:05:15.132 --> 00:05:16.951
So, yeah, be excited.
00:05:16.951 --> 00:05:28.173
You want to get into newer technology, you want to be able to reach customers in better and smarter ways, but at the same time, bring in a healthy level of skepticism, figure out where to land in that model.
00:05:28.173 --> 00:05:35.209
And then, of course, you need to meet your budgets, and that's always the hardest pieces.
00:05:35.209 --> 00:05:43.196
Where do I get the new toys while still maintaining the old toys and driving my skills to get there and still be?
00:05:43.216 --> 00:05:43.898
within budget.
00:05:43.898 --> 00:05:46.831
So we'll get to the pessimism soon.
00:05:46.831 --> 00:05:48.615
Let's get to the optimism first, right.
00:05:48.615 --> 00:05:51.208
So what are the new toys that we need to be thinking about?
00:05:51.208 --> 00:05:52.509
What are right?
00:05:52.509 --> 00:06:04.387
If I'm scaling out my operations and data and support systems for the applications that my organization are going to need to compete in this world, what do I got to do?
00:06:04.387 --> 00:06:04.968
What's hot?
00:06:05.769 --> 00:06:06.048
I think.
00:06:06.048 --> 00:06:14.300
So a few things I'm seeing from anywhere, from the fitness world through to even telecoms providers, is sensory devices.
00:06:14.300 --> 00:06:17.694
You know it's IoT plus, if you like.
00:06:17.694 --> 00:06:24.978
You want to be able to service your customer and give them better experience, but proactively, as opposed to just responsive to something.
00:06:24.978 --> 00:06:25.860
So I'll give you an example.
00:06:25.860 --> 00:06:32.595
You're a broadband customer 99, almost all the time You're the one calling in saying oh, it's down, I can't.
00:06:32.595 --> 00:06:33.790
You know, I can't get back online.
00:06:34.331 --> 00:06:55.776
So imagine if you had broadband piece that has other connectivity and other sensors, that recognizes that there's an issue in the area, create some sort of ticket in, you know, encourages some sort of communication to you whether it's to your phone or some other method, depending on all the devices you have and resolves the problem for you in in a succinct manner, at a low cross point.
00:06:55.776 --> 00:07:00.550
You know that that kind of thing is great and then you can go to the more trendy stuff.
00:07:00.550 --> 00:07:04.589
You know your fitbit type of idea or you know, not trying to sell fit, but but you get what I'm saying.
00:07:04.589 --> 00:07:05.976
You know your fitness devices.
00:07:05.976 --> 00:07:07.985
You've got a lot of century pieces there now.
00:07:08.026 --> 00:07:16.625
Those are getting exponentially better where they're tending more to health rather than obsession of looking at oh, you know how many sleeps did I have, how many steps I took.
00:07:16.625 --> 00:07:20.019
It's actually looking at oh, what's your actual energy consumption?
00:07:20.019 --> 00:07:23.951
What is your actionable impact on your body from a health perspective?
00:07:23.951 --> 00:07:25.976
Those are all developing significantly Now.
00:07:25.976 --> 00:07:26.237
Those.
00:07:26.237 --> 00:07:29.507
You could then extend that into things like insurance.
00:07:29.507 --> 00:07:32.850
You can extend it into life cover, all of these sort of things.
00:07:32.850 --> 00:07:45.641
There could be drivers that will help you either get a discounted insurance or help you in a medical sense of balancing between getting a discounted insurance and having the right medical piece to cover you on both ends.
00:07:50.055 --> 00:07:51.944
That's an ever-expanding universe, in my opinion.
00:07:51.944 --> 00:07:54.670
What kinds of tools are good for capturing this?
00:07:54.670 --> 00:08:06.288
I mean, and how does that connect with, say, a more kind of an industrial scenario where you've got tractors or you know other sensory data that isn't human, or maybe, maybe staff members that you're keeping track of, or stuff like that?
00:08:06.608 --> 00:08:12.713
so I was just thinking in terms of, just say, take agriculture, if you like, especially extensive agriculture.
00:08:12.713 --> 00:08:17.511
You know you've got gigantic water systems, you've got all sorts of stuff there.
00:08:17.511 --> 00:08:19.055
That's dealing with agriculture.
00:08:19.055 --> 00:08:23.173
Now, if you imagine, if you have some sort of vision technology, that's there.
00:08:23.173 --> 00:08:24.778
It's monitoring the health of your crop.
00:08:24.778 --> 00:08:31.630
It's also determining whether it needs to be watered, at what time of the day, when does it need to be fed, you know when's best to harvest.
00:08:31.630 --> 00:08:41.874
And it's not to drive down the need for people, it's more to drive up the accuracy and predictability of your productivity, of your farm.
00:08:41.874 --> 00:08:43.631
And that's an example.
00:08:44.924 --> 00:08:57.933
I think that ultimately, people are just coming up with such smart ideas to address what sometimes might seem like trivial challenges, and I always weigh up the trade-off in my mind.
00:08:57.933 --> 00:09:15.065
I remember doing some work in the Middle East years ago and we were trying to sell an accounting package for argument's sake and the key stakeholder I was dealing with said to me well, I can hire 1,000 people for less than the accounting package and they'll do a better job from his opinion.
00:09:15.065 --> 00:09:17.038
So we ended up in quite.
00:09:17.038 --> 00:09:24.333
In fact we had a good dinner and debated that about accuracy and all the rest of it versus just feeding 1,000 people.
00:09:24.333 --> 00:09:27.153
And you know you've got to take it in balance.
00:09:28.765 --> 00:09:30.212
So let's get real, though.
00:09:30.212 --> 00:09:47.227
What are the tools that infrastructure, hybrid infrastructure, architects and operations and technologists are really interested in, that are forward-looking enough, that are allowing us to start preparing our data centers and our operational plans to support this environment?
00:09:47.970 --> 00:09:56.591
So I think networking is at the heart of all connected devices and if you layer on top of that, you're looking at integration points.
00:09:56.591 --> 00:10:05.432
How do all these fancy tools integrate to whatever your brain portion of your technology is, where all the thinking gets done?
00:10:05.432 --> 00:10:07.025
Where does all the reporting get done?
00:10:07.025 --> 00:10:09.553
How do you manage resources, all those aspects?
00:10:09.553 --> 00:10:15.386
So it could be CRM all the way through to ERP type systems.
00:10:15.386 --> 00:10:27.615
The interaction point is your integration point and there again the choices are large but the availability of different systems is fairly large.
00:10:27.615 --> 00:10:37.052
But the core piece of it is security, really data management compliances, you know, and those are ever-growing.
00:10:37.052 --> 00:10:39.890
Even the people aspect of it.
00:10:39.890 --> 00:10:41.735
You know who services these things.
00:10:41.735 --> 00:10:49.452
I mean, if you've got a whole lot of connected devices out in the sticks, how do you make sure that you've got a trained engineer that can actually travel out there when things go wrong?
00:10:49.452 --> 00:10:53.275
I know it's not exactly tools, but there is such a plethora of those.
00:10:54.924 --> 00:10:57.687
Well, no, and it's actually just trying to, you know.
00:10:57.687 --> 00:11:00.169
I mean, I think we want to be able to figure out.
00:11:00.169 --> 00:11:01.610
What should I?
00:11:01.610 --> 00:11:05.835
How do I start getting ready for the world that's to come?
00:11:05.835 --> 00:11:14.142
Because I think that we're so overwhelmed with voices saying, oh, you're not going to need, you know, it or technical people anymore.
00:11:16.605 --> 00:11:18.307
To me it looks like we're going to need twice as many, yeah.
00:11:18.326 --> 00:11:44.787
And if you want to be a strategist in mind and you want to think of short, medium, long term and you want to think of how do I create something now that will bring some benefit now, maybe a lot of benefit, depending on your model, but you know that it keeps the door open for future pieces and again, to me, an integration model in your architecture is really important, something that's got a future to it and you may start off.
00:11:44.787 --> 00:11:56.274
I would always say start off small, elegant, succinct if you like in certain things, but think big where you might go and don't be afraid to experiment.
00:11:56.274 --> 00:11:57.226
I mean, you know.
00:11:57.226 --> 00:12:06.493
So that's why you start with small things, so that you're not spending too much on it and you're guaranteed to fail a few times while you're learning and, you know, while your teams are coming up to it.
00:12:06.493 --> 00:12:18.495
But once you've got that down pat, you then start looking at well, actually, okay, I've got 10 use cases now because we've been experimenting and people have been thinking about it, people in the business have seen some value from that.
00:12:18.495 --> 00:12:25.721
These use cases start coming through and then that will help you position a budget that will help you get to probably a better tool.
00:12:25.801 --> 00:12:35.086
So you know you'd probably start with quite an open market type thing you know, apogee or some other api type tool, and then later on you might go well.
00:12:35.086 --> 00:12:47.335
Actually you know our skill set's grown so much we can actually create our own model for it, but something that's particular to our industry, which you then might actually market out to the industry ultimately.
00:12:47.335 --> 00:12:48.529
So there's that route.
00:12:48.529 --> 00:12:59.484
Or you could say I still want to stay in the generic world and just build those integration models in a clever way that just repeatedly uses the generic piece rather than going too specific.
00:12:59.484 --> 00:13:07.037
And that's also again horses for courses, and we've seen pros and cons to that, that mindset for years.
00:13:07.037 --> 00:13:14.674
Whether someone goes bespoke or the say generic and cause even staying generic, you might find the tool gets deprecated.
00:13:15.376 --> 00:13:15.957
It's not uncommon.
00:13:15.957 --> 00:13:17.307
What's here?
00:13:17.307 --> 00:13:17.587
Okay?
00:13:17.587 --> 00:13:18.912
So let's just kind of back up for a second.
00:13:18.912 --> 00:13:21.067
When you go into a client, right?
00:13:21.067 --> 00:13:26.436
A client says, uh, I want you, matt, you're an amazing architect.
00:13:26.436 --> 00:13:29.508
What's your sort of 30 day?
00:13:29.508 --> 00:13:31.493
You know, how do I get a handle on?
00:13:31.493 --> 00:13:32.053
How do you get?
00:13:32.053 --> 00:13:33.437
How do you get a handle on?
00:13:33.437 --> 00:13:35.086
What are they doing?
00:13:35.086 --> 00:13:36.509
Why are they doing it?
00:13:36.509 --> 00:13:37.429
What they missing?
00:13:37.429 --> 00:13:38.552
Where do they need to go?
00:13:38.552 --> 00:13:44.539
Next, what's your personal method for infrastructure success, as it were?
00:13:44.539 --> 00:13:45.760
Right the recipe.
00:13:47.067 --> 00:13:50.335
I definitely start as far away from the technology as possible.
00:13:50.355 --> 00:14:00.294
I start with people Get to understand who are the stakeholders, who are the people involved, because it'll live or die on the people that are involved in any know any sort of program of work.
00:14:01.025 --> 00:14:33.950
Then understand the business strategy that they are trying to achieve, or whether they even have a business strategy that has a technical consideration, and then help work on that to find a, you know, to develop a roadmap that will both take into account the current skill sets, current people, current structures, and then a route to the future where they may want to go, and just showing some pragmatic value drops that you get as you go along the roadmap and you just got to remember.
00:14:33.950 --> 00:14:43.977
So, whatever roadmap you put in place, you know if they're looking at five or 10 year strategy, but they're an annual budget driven company that roadmap is going to pivot multiple times.
00:14:43.977 --> 00:14:52.571
So ensure that you're building value on that, that either they can expand on the value or you could pivot from that without ruining everything.
00:14:52.571 --> 00:14:59.331
And there's that experience and it might sound counterintuitive initially but it just makes sense.
00:14:59.331 --> 00:15:08.571
I've met a lot of people who wedded to their roadmap no matter what, even though everything's changing all the time and I think we've all seen those white elephants.
00:15:10.296 --> 00:15:26.517
So I guess the next question I have for you, the last question for this podcast, is what is AI going to do to infrastructure and especially, and maybe with a bent towards sustainability?
00:15:26.517 --> 00:15:37.542
We just came from our sustainable architectures event and I I know you were there and even they're sick or fighting sickness, but what is AI gonna do to our infrastructures?
00:15:37.783 --> 00:15:39.330
if I to put my my childhood hat on, if you like.
00:15:39.330 --> 00:15:41.745
So a put my childhood hat on, if you like.
00:15:41.745 --> 00:15:44.374
So a bit of imagination and a dream, if you like.
00:15:44.374 --> 00:16:10.854
I would love to see AI getting to a point where it's integrated with all abstraction that you have for infrastructure and it really handles your L1, l2 considerations from an operational perspective, including forecasting and an expansion of resources if you need to, especially, you know, obviously, cloud lends itself well to that, but you would also be able to do it on-prem if you had the right kind of measures in place.
00:16:11.426 --> 00:16:12.772
That's where I'd love to see where it goes.
00:16:12.772 --> 00:16:19.734
I'd still think that even with that and you're not dealing with personal data or anything you would still need a bit of a human in the loop while that's learning.
00:16:19.734 --> 00:16:24.096
But I think that's a bridge somewhere between RPA, ml and AI.
00:16:24.096 --> 00:16:28.253
Everyone will call it the same thing, just from a marketing perspective.
00:16:28.253 --> 00:16:29.636
But you get where I'm going with that.
00:16:29.636 --> 00:16:44.711
I honestly think it will be more involved in the innovation of how technology goes, whether it's higher density, smarter way of doing storage, intelligent networks beyond just a software-defined level, intelligent threat management.
00:16:46.047 --> 00:17:00.894
I think these things will, and then I honestly hope that we'll see AI coming into FinOps so we get a much better cost control measure and better reporting than that, than just a few opinions that drive some architecture around it.
00:17:00.894 --> 00:17:14.417
I think it needs to get a bit more and and with time, with more data, you'll get better outputs, you know over time, and I think that will really help so I really appreciate you being on the show.
00:17:15.019 --> 00:17:25.880
For those of you that are interested in chatting with matt, matt is going to do a lot more than just show up on a couple of my podcasts, even much as I enjoy these talks.
00:17:25.880 --> 00:17:31.036
He's going to be joining us in this location.
00:17:31.036 --> 00:17:40.316
This is I will put that on your podcast and in the YouTube comments, et cetera.
00:17:40.316 --> 00:17:53.367
That is our mighty network for hybrid infrastructure, and we're going to be bringing speakers like Matt, a speaker and podcast events, roundtables and a whole bunch more.
00:17:53.367 --> 00:17:55.349
It's a free community.
00:17:55.349 --> 00:18:03.040
It is as hype free as we can get it, as you can tell, and Matt's going to be there.
00:18:03.040 --> 00:18:05.871
We're going to be doing some publications there, et cetera.
00:18:05.931 --> 00:18:11.951
So I really, really urge all of my listeners to go sign up on that location.
00:18:11.951 --> 00:18:14.857
Join us in the ISO architecture community.
00:18:14.857 --> 00:18:19.634
Help us change the world right, Matt.
00:18:19.634 --> 00:18:21.960
Thank you so much for joining us today.
00:18:21.960 --> 00:18:24.907
Uh, as always, I love talking to you.
00:18:24.907 --> 00:18:26.231
This has been great.
00:18:26.231 --> 00:18:33.193
I like doing these little tidbit sections and I know you were a bit nervous ahead of time so, yeah, just totally hit it out of the park.
00:18:33.233 --> 00:18:34.396
No problem, pleasure.
00:18:34.396 --> 00:18:35.205
Thank you, paul, all right.
00:18:35.226 --> 00:18:49.038
Looking forward to the next and the next podcast I will be launching uh, we, thank you, paul, all right everybody join me, and the next podcast I will be launching we will be talking about technical debt with a special guest.
00:18:49.038 --> 00:18:57.592
So join or click on my Buzzsprout lead for the argument or follow the channel on YouTube.
00:18:57.592 --> 00:18:58.894
Thanks so much, matt.
00:18:59.416 --> 00:19:00.579
Thank you, take care.