Speaker 1: 0:02

Welcome to Uplink, where we explore the world of digital infrastructure, uncovering the technology fueling AI and cloud innovation, with the leaders making it happen. All right, well, Jonathan, thank you for joining us. Why don't you just quick introduction for the folks listening and then we'll get into the conversation, phil, thanks for having me.

Speaker 2: 0:34

Michael, so Jonathan's staff. Everybody calls me Jono and currently I'm the head of private cloud at Macquarie Cloud Services, part of Macquarie Technology Group, so my role encompasses lots of Services, part of Macquarie Technology Group, so my role encompasses lots of things, lots of things, but primarily what I do is just help customers find a safe and cost-effective place to run applications that are important to their business.

Speaker 1: 1:00

Yeah, and it's a pretty hot space and it has been a hot space for a while.

Speaker 2: 1:05

Yeah, and it's a pretty hot space and it has been a hot space for a while. Yeah, look, I think the infrastructure as a service game has been hot for a long time. Depending on what reports you read around, the growth of cloud and infrastructure as a service certainly in the Australian market, compound annual growth rate succeeding 30% for years and years and it seems to be kind of recession proof or it just keeps going. But I know that that's different in different market verticals and all the rest, and I think underneath that there's a lot of different drivers, a lot of different drivers. But one thing that's definitely for sure is that everybody's looking for efficiency, right, and there's a funny rule. It's kind of the more efficient this technology gets, the more of it people seem to consume, or businesses or the market seems to need to consume.

Speaker 1: 1:58

So that's been sort of something interesting to watch over the last few years, and so you're in a different space to the traditional. I guess the cloud providers have been going at this for some time. There was this sort of big swing away from private data centers and then the cloud is going to solve all these problems and we started seeing massive migrations into the cloud. I think originally it was a lot about. It was just so much easier we could develop and move faster. But I think what happened was we saw all these large companies just move a workload that was sitting on-premise and just sort of probably not repurpose but literally migrate into the clouds. And then so we're seeing globally all these different trends between some folks are choosing to stay on-premise, some are moving some services to multi-cloud providers who can offer different either cost options or different specific services for a particular function. How do you guys see that playing out? Because I know you've built your own private cloud. It'd be really interesting to understand sort of the dynamic between all of those factors.

Speaker 2: 3:00

Yeah, sure, there's a lot in that, I'd say.

Speaker 2: 3:04

Probably the first point Wari Cloud Services has been in sort of the infrastructure as a service hosting game for over 15 years about 15 years now and so we've always had a pretty substantial, what he called it hosting so quite a while and has seen you know, varying sort of ebbs and flows of demand, but obviously one thing that's true is that it's grown year on year. I think the hyperscalers coming to Australia was definitely disruptive for that space sort of the private hosting space and there was a huge rush to get in. And I think you're right, something that I witnessed in my career even before Macquarie, hyperscale cloud was a pretty hot proposition. There was a huge rush for IT organizations to get their applications in, whether they were fit for purpose for that cloud or not, and a lot of that was the promise of utility-type commercial models, attractive rebates, hype, like all of these things came together.

Speaker 2: 4:09

I think, though, as those technologies become more efficient and there's more widgets and things to consume, one thing that's been true is that the cost, the OPEX, associated with running your business on those platforms has also continued to grow, not necessarily in line with revenues in a business, so it starts becoming a huge cost and then obviously starts to attract some scrutiny from executives and boards and shareholders to say, well, hang on, how do we sort of arrest some of the growth in our costs in these platforms? One of the things that has started to become popular is, I suppose, the notion of hybrid. Really, to me, what that's about is just putting the right workload in the right place.

Speaker 2: 4:58

And I have a view of this tension that IT leaders are in at the moment. There's the efficiency and ease of use and deploying new things at scale and time to market. Those things are attractive high-scale cloud use cases for development teams, infrastructure teams. But really, when it comes to cost management, if you're all in on any one of those platforms, you're limited in the number of levers that you have to control your costs. So if someone was to come to you tomorrow as a leader in that space and say, listen, everything you're doing is great, but we need you to find 20%, you got to save 20%. There's a few options available to you. You can spend more and try and attract some kind of volume break discount. So all good Less popular, you can enter into some.

Speaker 1: 5:53

I'd pitch to go to the CFO with you.

Speaker 2: 5:56

I don't know if someone brought that to you, michael. Okay, maybe that could work. Or you could enter into some kind of take-or-pay arrangement where you make a commitment or you make a commitment for the amount of these services that you'll consume over a period of time.

Speaker 2: 6:20

So that's more of a risk-sharing model. You de-risk your provider and they will reward you with some kind of discount so you could improve your rates. Provisions, these kinds of things are popular, so you can do that. But that comes with its own challenges. So you can spend more, you can buy more stuff, you can enter into take-all pay agreements or you can turn some stuff off. That's pretty much it when you think about it. Now, if someone comes to you and says, well, we've got to find 20 and you're already doing those things because they're pretty standard finops, you know cloud cost management disciplines. Well then, what do you do? Yeah, right, what do you do? And really the answer is you you need to look at other platforms and more cost-effective options for you to be housing those application workloads or those data sets.

Speaker 1: 7:13

So on the cost side, the challenge is, I think you're right that since sort of that post-COVID era where every company on the planet was really focused on cost in every aspect, I think most companies got smart and they figured out how to optimize and turn off services that were inside clouds that weren't being used. There's all these tools and processes to go and sort of optimize that down and so, yeah, maybe they've already got that 20% saving putting all that effort in. And the question is the same question comes back down which is solve for this? Yet again, your point was interesting. I didn't think about it, but it's like if your company's growing at, say, 10 or 15 or 20%, but your cloud costs are growing at 40% or 30% or whatever it may be, you end up in this position where you're sort of having to have a conversation with your CFO and CEO, et cetera. So it's like, how do you solve for it? And, by the way, this isn't anti-cloud.

Speaker 1: 8:08

I think spends are just becoming out of control in different ways. It's a bit like all the subscriptions I think I have. I have subscriptions all over the place. Every time I sign up with something on my iPhone, I've got another $10, $15 subscription and by the end of the year you realize you're paying a fortune for all these different services. You probably didn't even realize you had Things get a little bit out of control.

Speaker 1: 8:28

What I like about what you guys do is you look at the entire picture and start to say, well, there's certain things that make sense here. There's certain things that make sense there. You can optimize things in the cloud. You can also say there's certain functions that actually are hybrid, and I also know you extend clouds as well. So I think Azure is a good example. I think I know you guys are huge partners with Azure and also with the private cloud components. So, look, there's lots of pieces to that puzzle and that's why I think you're seeing it all play out. So what are you seeing the best results for customers? So for people listening, what are some examples of some of your customers that have solved some of these equations and how did you go about it?

Speaker 2: 9:09

Yeah, great question. So we've sort of published some recent case studies on this, but I would group customers that are doing this really well into a couple of categories. I think the ISV vertical, so independent software vendors that's sort of a discipline in its own, Highly capable, lots of cloud management maturity, but need help.

Speaker 1: 9:32

Describe what's an ISV.

Speaker 2: 9:35

Yeah, an independent software vendor. So a SaaS business, a business that makes software? Yes, yeah, a business that makes software. They're in the business of monetizing their IP. They're great developers not necessarily great infrastructure people like the plumbing and the infrastructure right, and the attractive thing for those organizations is obviously the way that the infrastructure is obfuscated in a hyperscale environment for developers, right. But when it comes time to actually then think about infrastructure, because we're optimized within our current cloud, but when you're doing that in kind of an echo chamber, there's only so so far you can go we need to figure out well, how do I actually run?

Speaker 2: 10:16

how do I run this application across multiple clouds? How do I maybe repatriate some of the data onto more cost-effective storage options? How do I take some of the data onto more cost-effective storage options? How do I take some of the workload out of, like the base load on the infrastructure out of this cloud because I'm paying a premium for it here and move it into another environment over here but leave my layer there? These are all infrastructure conversations. So we've had some good success good success helping organizations like that.

Speaker 2: 10:45

There's a public case study around a company called Collaboro, great Australian business doing interesting stuff around digital asset management. Probably a big concern for them was in the digital asset management game. They're in the game of generating and protecting lots of data, right, petabytes of data. Now, as your business grows, so does the amount of data that you're ingesting. But where they were?

Speaker 2: 11:10

The public cloud that they were operating in, or the hyperscale cloud they're operating in. It was good when they started, but now, as they scale, the OPEX of actually housing that data for their customers was starting to outpace their revenues. Right, and it's margin eroding right, and it's margin eroding right, of course. So then, how do you actually stop that and arrest that and get it back into land so you can actually grow your company without your costs eating into your margins? And what we identified there was by repatriating certain data sets out of that environment, and being able to hedge on the price of storage actually helped them arrest that. The way to do that, though, was to obviously have to connect multiple clouds to that environment. We need to connect other infrastructure, actually out of our private cloud systems, up into their hyperscale environment in a seamless way, and that sort of comes down to how do you bring true hybrid cloud to life? It's about a consistent operational experience across the top.

Speaker 1: 12:13

And you're sort of going into a space where we live which is that network element. We spend our lives connecting everything together as much as we can and giving customers as much choice. It's sort of funny. It's like the network is always. It's like oxygen people forget it's in the room, but without it you quickly, you know, you run out of air and you die. The OurSpace is enabling you guys to sort of connect between all those clouds and provide those services to the customers and give them choice beyond what was probably a lock into one or two particular folks.

Speaker 2: 12:48

That's exactly right To go a little bit deeper on that. There were two things that were really attractive about partnering with Megaport to help that customer. The first one was the speed with which we can deploy the connectivity right. Because you're lit up in our data centers, it's really easy for my engineers to turn it on right. We're not waiting for Rebuilt 60 seconds bang bang.

Speaker 2: 13:09

Yeah, we're not waiting for like six weeks and then we're embargoes and things and stuff with telcos and all the the stuff. The way it used to be right, that was the way it used to be so that in itself is we say we're very, we say we're very un-telco yeah, yeah, I think we we actually say that too.

Speaker 2: 13:26

Right, so it's a quarry, so yeah. So I think that's an important factor, because time is money. Right, the sooner you can get a customer out, save them some money, the better, so that's great. And those savings can subsidize the cost of change.

Speaker 2: 13:41

So, speed is everything. And the second thing was there's a cheeky barrier. There's little commercial barriers everywhere. So I talked about organizations who've gone all in on hyperscale cloud. They've got big balance sheets to figure out how to get you to spend more than you're comfortable with spending. They've got this down to a finite and I think one of the things is egress costs.

Speaker 1: 14:05

Yes, and ingress. Let me just pause you for a sec. Yeah, so a little wardrobe change there for the audio. We were getting into the conversation around sort of some of the what I would say hidden costs inside cloud that you sort of creep up on you and you sort of don't realize what's happening. Egress, I think you were mentioning as a really good example of that, and I'll chat about some of the new products we've just launched in that gateway as well, which is addressing the same thing, but I'd love your perspective on that.

Speaker 2: 14:37

Yeah, sure, one of the things that's important with managing egress is the speed with which you can extract your data.

Speaker 1: 14:46

Actually what is?

Speaker 2: 14:46

egress. It's data coming out, right, data leaving the cloud. It's out of leaving one place to go to another place. So the data leaving, or, as a hyperscaler might view it, you know revenue churn, um, they're kind of the same thing, right, so you'll pay a fee for that, right, and if you've got a lot of data, it can be a pretty big cost. You know in the tens of thousands of dollars. So even though you might be able to cost, you know in the tens of thousands of dollars, so even though you might be able to go, you know, to another cloud, say 20 on your per gig rate for that storage tier, getting it out might cost you a couple of hundred grand that you don't have.

Speaker 1: 15:23

So the cost of change can be really high, especially when you're dealing with light phones, with data, and if you live in both and you're constantly sending traffic back and forward, obviously that becomes an issue as well.

Speaker 2: 15:35

Yeah, it can be. So it's something that's really important when you're considering a hybrid cloud strategy is the amount of chatter back and forth between application groups. One way to think of it is you've got to understand how your apps and your workloads communicate, and the idea is to move them into bubbles and reduce the amount of chatter across between clouds to reduce your egress charges exposure. So there are kind of two scenarios. One is a repatriation scenario where you're getting the data out and there's a barrier to do that, and the other scenario is okay, we're in a hybrid cloud state, a future state. Architecting that correctly to understand that dynamic of the chattiness between the application workloads and reducing and mitigating that also is a consideration in the design of those environments.

Speaker 2: 16:26

A thing that was really interesting with Megaport, specifically around the Collabro use case, was we were able to turn on a high bandwidth connection very quickly, which enabled us to move more data out in a shorter amount of time. And this is where there's a little bit of a trick, because Megaport's got POPs in the right locations. If you pop that data out of certain hyperscale providers and it lands in the right location first, as a first hop, there's big discounts on the egress fees Interesting. So where the data comes out, the physical location. So where the data comes out, the physical location of where the data comes out is really important and megaport can help help with that right. So you might have it where, okay, the data's coming coming out and it's landing here, and then we're moving it from here to here and that move there's no egress fees because it might be over a private dark fiber that, say, macquarie's deployed between data centers, or over another Megaport link, as an example.

Speaker 2: 17:38

So that's a consideration. And then being able to do that in a short space of time. So imagine you had, say, a private connection with your hyperscaler in, say, an Azure scenario. That's an express route. And then the type of express route that you purchase has implications on your egress charges, right, and also the rate limiting of the bandwidth, the amount of data that you're able to move out at any one time. So if you want a high bandwidth link, it's going to cost you more, and the longer that that link is in place, the more it's going to cost you for every month. So if you can reduce the amount of time it takes to ship data out, from three months to one, you're going to reduce the cost of that express route uplift by two thirds.

Speaker 1: 18:31

And there's a lot of companies that punch it straight out through internet, which is going to get a huge egress cost. So that is where it goes, that's right.

Speaker 2: 18:39

So being really clever and discerning around and creative around. Well, how do we reduce the amount of time it takes to get the data out? We need more bandwidth so you can do more data at once. So, tick, how do we make sure that we're landing it in the most cost-effective place? First, right, wherever the hyperscale has decided it's important, like they'll charge less for that data to land. Yeah, help with that tick. And how do we turn this on really quickly so we're not still paying? The high rate of storage or the opportunity cost, right is high that the longer that it that it takes. And how do we actually do this project really quickly? And so those dynamics are things that we think about at Macquarie and then, and then I step back and I actually Megaport's kind of underpinning all of that as like the silent hero enabling our engineers to actually do some of this cool stuff. So, yeah, really really interesting nuances there with egress.

Speaker 1: 19:35

Well, we've got 230 cloud on-ramps globally, so we have all these different locations that we've landed in to try and solve for these different equations. And you're right, it's a little tricky to figure it all out. You need some smarts to come together and design it, and each cloud provider is slightly different and, to your point, like each location is slightly different, you have different zones. We've even found that you're sort of hair pitting from one zone to another, even though you're in the zone, and then you're getting charged. There's all sorts of there's a lot of science in it and it's not. It's not as straightforward, which is why having someone like yourselves come in can probably save a huge amount of cost to get at least and also to solve that puzzle.

Speaker 1: 20:14

Moving forward, I wanted to ask you around we're talking about. I think the world's pivoted so hard towards AI. Everyone's like I'm going to do this GPU as a service, but I think people are also forgetting that a huge part of the business is still run on what we would say is traditional compute, traditional storage, network capabilities and so forth, and that's the big problem customers still have to solve. And then we've got this AI thing. So how are you seeing AI play out? Are you seeing your customers? Do they need GPUs? Are they trying to wait and see? Are they pushing into SaaS providers to do some training? I don't know how. Are you seeing it play out? And I'm cognizant you've got a big presence with government, so there's probably some stuff you can't talk about, but maybe, more generally, what you can share would be interesting.

Speaker 2: 21:03

Yeah great. How are we seeing it play out? I'd say there's in the early stages of what I think is going to be a pretty steep demand curve. So right now we've got some early adopter type customers who are curious and by curious I mean thinking about proof of concepts for very specific niche use cases where there's a clear path to return on investment in their business. What does that mean from an infrastructure perspective or from a quarry? There's two areas that we're getting drawn into. The first one, actually before the infrastructure, is where is our data? Who's responsible for our data? What are the security implications of doing something different with this data? Are we going to inadvertently create new attack vectors?

Speaker 1: 21:56

Are we sending it to China?

Speaker 2: 21:58

We get drawn into a lot of those conversations early on, as in the governance frameworks and how to do it safely. Governance frameworks and and how to do it safely and beyond that, there's a lot of early interest in, I would say, low code, no code capability, that's what does that?

Speaker 2: 22:14

mean, yeah, so organizations evaluating you know things like microsoft fabric as an example. If you think about how you you know go back a few years and it was a game changer to be able you didn't have to be an excel guru to use power bi and drag and drop and create dashboards. It's kind of like low code, right yeah, so imagine so a lot of interest in that similar type of way of working.

Speaker 1: 22:39

So take it so like an example, would be like dumping a whole heap of your financial data into, say, an ai platform and then being able to, just, you know, create graphs, create insight, or and just ask it versus like I need to write a question yeah, yeah, yeah, that's right like uh, you might say what were my sales numbers, like you know, last quarter and give me a projection for revenue or net profit over the next three years based on this data set.

Speaker 2: 23:07

Like these kinds of things are what I'm seeing. A lot of emerging use cases Like the Australian market's interesting because the majority of the market is like mid-market or North America would call it SMB, so they don't have huge IT teams or you know, armies of analysts or these so low-code's super interesting. You know how can I do something with this technology without massive investments in my payroll?

Speaker 1: 23:31

Or CapEx, which is like Exactly, yeah, no one's got the money to buy a whole bunch of NVIDIA chips. Hyperscalers obviously do, and there's a few folks that are building the Terminator or whatever it may be that are punching into that. But yeah, traditional companies can't really take that investment and also, if they did make the investment, it's changing so quickly. I think that's right.

Speaker 2: 23:51

They might have a wrong chip yeah, yeah, they might. And then on the infrastructure as a service side and more. I would say, specifically, in the business, I'm responsible for a lot of conversations around high-performance compute and storage and optimizing our costs and getting our house in order so we can get ready to scale this capability, high-performance compute specifically, I would say most of the demand as I see it right now is coming from research-type organizations higher education, sector, medical research, like those traditional verticals that you would expect would be early adopters or have more maturity, I suppose, where they're able to say you know what I need, I need this kind of high performance compute capability that they're very that they already know what they need right and is that high performance compute?

Speaker 1: 24:45

it's, that's not gpus, that's actually incredibly fast compute traditional computers. Or is it cross-intuitive?

Speaker 2: 24:52

To me, I think it's both. Not every AI model actually needs a GPU, so that's probably a common misconception. You can do it on really fast traditional processing. I think I'm seeing mixtures of both and the other one is the speed of the storage. Can you guarantee the performance of the storage, iops et cetera, and can you guarantee that at scale? You know how quickly will it take you to turn more on? What's the price per gig? So everyone's thinking about how do I optimize for costs, how do I think about my data and where it's stored, and how do I secure it and get ready to do something clever with it? And what does high performance computing mean for me? I don't want to buy more than I need. How do I cut through the hype? Getting drawn into a lot of those type of conversations, yeah.

Speaker 1: 25:41

So it is such a frothy space. It's changing rapidly, like we just experienced when DeepSeat came out and the market sort of just went whoa, but it was only two years earlier that sort of Chezip Chee came out. So it's just, all these things are moving really, really quickly. It's hard to keep up with it. To your point, customers, I think, are getting a bit of fatigue, as in they're just. People are thinking well, we need to be here. I'm being told by everyone, we have to be here. I'm being told by everyone, we have to be there.

Speaker 1: 26:07

Boards are saying what are you doing from an AI perspective? Ceos are thinking well, push that to the team. The team are like we've got Copilot, there's all these different sort of pieces coming out. But working with someone like yourself who's seen so many different customer examples of that, I think that's where the value that you bring you can share. What success is happening with different companies for different styles of workloads, for different, I guess, examples that they're trying to run and maybe just talk through the value that you, you see, you guys bring into your customer base yeah, cool.

Speaker 2: 26:41

I think, as a managed services provider, the way that we create value is to help customers solve problems, right?

Speaker 2: 26:49

it's actually that old trick, yeah actually agree, you know, and and to do some good business we've got to be, you know good about exchanging those solutions for money. So I kind of like try and keep it very simple and the problems that we solve for we're very discerning around that. So Macquarie's not a, I'd say, a traditional managed services provider. We don't, we don't try and be everywhere, we don't try and solve for lots of things.

Speaker 1: 27:15

It's actually very focused and I think that's the value in that very yeah, absolutely targeted, and the value that you provide is going to be immense. In the specific thing you're focused, I think that's right.

Speaker 2: 27:27

You know we're not going to come and fix your laptop or reset a password or all that sort of stuff. We're actually just we've gone very deep deliberately on solving for cost and risk for your mission-critical application workloads.

Speaker 2: 27:45

So if you're looking for a safe, cost-effective place to run applications that are important to your business. That's where Macquarie has gone very deep and that can span across public cloud. We've gone very deep with Microsoft Actually, I believe it's possible to be excellent at all of the hyperscale clouds in terms of addressing these problems. So we've deliberately gone very deep. Right, if you're looking for a private cloud solution, we've gone very deep in an infrastructure as a service platform that we've built for purpose around that, and we've gone very deep on cybersecurity managed detection and response XD. So you can now imagine if we're responsible for the management of your mission-critical application infrastructure across one of those two environments or both mostly both for our customers from a hybrid standpoint and then we're responsible for the cybersecurity around those environments. Yeah, perfect, that's actually it, so risk across mission-critical apps.

Speaker 2: 28:47

I love that and you perfect. That's actually it. So risk across mission crew collapse.

Speaker 1: 28:50

I love that and you've got your own data center actually a lot of data centers that you own and have built yourselves. You've got the infrastructure that you've sort of perfected the art of for over 15 years a lot longer than I think most. You've got the expertise, like you said, deep with Microsoft. It's not easy. This is one of the things that a lot of our solutions architects spend a lot of time helping our customers with. Just the networking requirements between the clouds are very different and how you show up and connect in the expectation for five, nines and all the different reliability that Microsoft will give. You need dual and redundant, you need different locations. There's all these rules, and so setting it up is actually not easy to your point. Even when we're massive in the US and even the customers over there, when they're large, just don't have the expertise and the skillset. So we spent a lot of time just helping on the network side. I can imagine when you step into the cloud there's just an enormous plethora of things that you can solve for. So it's awesome that you guys are around.

Speaker 1: 29:47

We appreciate the partnership. We know you've solved some of certainly Australia's largest challenges from a government perspective that a lot of the folks just wouldn't even know about, particularly from a cyber angle. We're grateful that you've used Megaport in many of those cases, and so we feel excited that we're actually helping make a difference here for citizens in the country. That we're actually helping make a difference here for citizens in the country, yeah, but we also see the impact you're making from not just we're just trying to deliver everything on our little platform on premise, and everyone else can do whatever they're doing. You're actually a hybrid between all of them, so you're going to be able to add the value to the customer where it makes sense.

Speaker 2: 30:23

So I appreciate you coming on. Yeah, all good, michael. Yeah, exposure to lots of different projects and giving the market choice around how to solve for these problems. Costs and risks on everybody's mind and I think choice is the biggest factor for success and I know that's what Meg is about as well, choice with safety and resiliency, ie ensuring that everything remains incredibly protected and safe.

Speaker 1: 30:51

The uptime and the speed and the uh and the service that you're providing is similar, if not better, uh. And then, to your point, you solve the cost problem without creating any other of those challenges, which is the hard part, and then you've got, like you said, you've got a business because someone's interested in reducing their costs, getting a better service. And that is how companies are formed. Yeah, it's solving a customer problem and doing that in a cost-effective way. We love it. Yeah, trying to keep it simple yeah.

Speaker 2: 31:19

That's great Thanks, michael. Thanks.

Speaker 1: 31:20

Jono Cheers.