Voices of Video
Explore the inner workings of video technology with Voices of Video: Inside the Tech. This podcast gathers industry experts and innovators to examine every facet of video technology, from decoding and encoding processes to the latest advancements in hardware versus software processing and codecs. Alongside these technical insights, we dive into practical techniques, emerging trends, and industry-shaping facts that define the future of video.
Ideal for engineers, developers, and tech enthusiasts, each episode offers hands-on advice and the in-depth knowledge you need to excel in today’s fast-evolving video landscape. Join us to master the tools, technologies, and trends driving the future of digital video.
Voices of Video
Efficiency, Economics, and Innovation - Transforming Video Architectures
The economics of video delivery are changing dramatically, forcing media companies to rethink their entire approach to content distribution architecture. In this fascinating roundtable discussion, leaders from Akamai, Scalstrm, and Arcadian reveal how the push for efficiency is reshaping video workflows across the industry.
As streaming platforms expand globally, they're discovering that delivery strategies that worked in established markets don't necessarily translate to new territories. Our experts explain how companies are bringing technical expertise in-house and carefully balancing where different workloads should run. The conversation explores why some processing makes more sense on-premises while other functions benefit from cloud flexibility, creating increasingly sophisticated hybrid deployments.
Power consumption emerges as a surprisingly central theme throughout the discussion. With electricity representing the largest operational expense for many providers and environmental concerns growing, our panelists share how specialized hardware like video processing units (VPUs) are dramatically improving encoding efficiency. You'll learn how these advancements are opening new possibilities for processing video at scale without breaking the bank.
The most thought-provoking segment comes when our experts discuss how AI-generated content is fundamentally changing video creation itself. With an estimated 50% of short-form video already being created without traditional encoding processes, the panel explores both the threats and opportunities this represents for everyone in the video delivery ecosystem.
Whether you're architecting a major streaming platform or managing video workflows for a content creator, this conversation offers valuable insights into the future of media delivery. Don't miss the practical advice on how to start transforming your own infrastructure, even when constrained by legacy systems or existing vendor relationships.
Stay tuned for more in-depth insights on video technology, trends, and practical applications. Subscribe to Voices of Video: Inside the Tech for exclusive, hands-on knowledge from the experts. For more resources, visit Voices of Video.
voices of video voices of video.
Mark Donnigan, NETINT:The voices of video voices of video well, welcome to another super exciting episode of voices of video. So we are now T-minus about five weeks five and a half weeks to IBC and the excitement is building. I am joined by three amazing companies and today we're going to have a really, really fabulous conversation. You're going to want to stay to the end and hear everything that our group has to say, both about what they're going to be doing at IBC, what they're going to be showing, but, more importantly, what they're really excited about Right now in the industry. We're going to be talking about some super exciting trends. There's probably going to be, you know, some surprises, so you definitely want to hang out, you want to listen and, as always, we thank you for hanging out and listening to Voices of Video. So, without further ado, I'd like to introduce our panel and, dominic, why don't we start with you? Introduce yourself and the company and what you do.
Dominique Vosters, Scalstrm:Thanks for inviting us again in the voices of video. Always good to be part of it. So I'm working Skalstrm. A company initially developed an origin packager mainly for tier one operators telco now also for broadcasters and then tier two operators and content owners. During the road we've developed more and more products from Origin Shield, cdn, thought Transcoding, just-in-time Transcoding and Live Transcoder, and there we met kind of run into NetInt, and that's, I think, why we're on Forces of Video.
Mark Donnigan, NETINT:So happy to be part of it. Awesome, awesome Joe.
Joe Waltzer, Arcadian:Yeah, thanks, Mark, for having me. I'm Joe Waltzer. I'm CEO and founder of Arcadian. We're a services company. We help people in the media space. So content creators we help them reach their customers in sort of efficient, exciting, technology-focused ways. So thanks for having me. I'm looking forward to the discussion.
Will Law, Akamai:That's great to have you and Will. Hey, mark, and likewise thanks for having us on. So I work at Akamai. I think people traditionally know Akamai as a CDN, but today we're a cybersecurity company, a compute vendor and a CDN, with the CDNs only about a third of our business nowadays, so we really evolved to be larger, but we still use for all three of those we use our very large distributed nature to be a great cybersecurity vendor and a compute vendor. As well as delivering content. I work in the cloud technology group. I specialize on the media side. I also work at IETF, w3c, svta and the standards orgs working on different standards for inter-off within the industry.
Mark Donnigan, NETINT:Yeah, it's very important work that you all are doing, so thank you for joining us. So, you know, I think I'd like to start our conversation here. It's easy, you know, to sort of go deep into technologies, into our products, into the platforms, and that's certainly a relevant discussion, right. But let's up-level and let's start with talking about why media companies are transforming their architectures and, joe, I think you've got a really good perspective on this based on what you're doing for large studios, large content owners, vertically integrated media companies. You know who are really under pressure right now to diversify their revenue, grow their revenues. So, you know, tell us, you know what are they building, what are they having you build, you know what are they thinking about in order to, you know, to continue to grow their business and compete.
Joe Waltzer, Arcadian:Yeah, it's a great question. It's an interesting time, I would say One of the big questions we get, one of the big asks, is how do we reach more customers right? So I think a lot of people feel like we've sort of solved, you know, that that first streaming thing. Nefas, came out and all the other clones came out and those are all fairly well standardized. And now we're seeing to be and certainly big players sort of broaden the reach. And now, as the cities are sort of seeing all this growth and seeing all of these opportunities as they expand new markets, expand new countries and territories, they're realizing that their assumptions about how to deliver to those users isn't the same, as you know, in the States and in places like Europe. So they're trying to find ways to be really efficient and really smart with how they deploy their content and sort of stepping away from these older school models where they just outsource everything and just worry about making content. So we've seen a lot more companies bring the tech in-house, bringing people who know and understand how to deliver content to users and do things like that, and we're getting a lot of questions about how do we save money and how do we deliver things as efficiently as possible. How do we reach those users in the countries that don't have our content yet, and we want them to have our content. So that's probably our biggest discussions.
Joe Waltzer, Arcadian:That's probably the biggest thing we deal with now is the best way to get this content. We're talking usually about bad catalog content. Right, I think everyone understands how to stream Spider-Man and Superman and all those those things. But when you start digging into your catalog and you have these, this content that isn't quite as valuable, doesn't quite have the same broad reach, that you have to find more specific uses. You really have to be good about your margins and your efficiencies and how you give that to customers. You can't really waste. You just can't waste any money there because the margins go away. So a lot of questions about how do we really get really efficient, really dig down and get this content in everyone's hands and make some money.
Mark Donnigan, NETINT:Yeah, absolutely, you know, will. I think Akamai probably has some interesting experience you know to bring here, because transforming, as you said, at this point I think it's a third of your revenue is CDN, which is pretty remarkable because maybe there's some listeners who still think of Akamai as like maybe two thirds would be CDN or even a higher percentage is like maybe two thirds would be CDN or even a higher percentage. So, as you have, you know, expanded your offering, especially around the connected cloud. You know what are, what are media companies coming to you for? You're still carrying a lot of traffic. You know one third of Akamai's revenue is still a big number, so it's still a lot of traffic. But how does this look for Akamai?
Will Law, Akamai:Yeah, I mean you should say why should Akamai expand? Firstly, I think I get asked that question all the time and a lot of it is to do what Joe just said. Everyone's looking for efficiencies. If people haven't noticed, delivery has become commoditized. It's no longer special, right, it's priced per gigabyte. There's not a lot of bells or whistles or features and what you've seen is a huge contraction in the number of CDN vendors. Three of the largest six have gone bankrupt in the last three years.
Will Law, Akamai:That's how tight the market is in the CDN sites. You have to be incredibly efficient in delivering video. There's more and more volume of video and the marginal cost of each new video deliver decreases, right, there's just a flood of it. So CDNs have to become extremely efficient and as a result of that, the growth rate's not high. So Akamai says well, how can we naturally leverage the assets we have, which is a highly distributed network? Well, video has to come from somewhere, it has to come from an origin. So why don't we go vertically into the compute and origin space? And then why don't we protect the content on the outside, which is crime, unfortunately, is the only thing growing faster than the internet. So that explains sort of the expansion.
Will Law, Akamai:What we see for media workloads on compute and these are the fastest growing part of our business is media on our compute servers, because we have a very high egress capacity combined with this edge-based compute, and that's the sweet spot for media, right? Media is a heavy workload. You don't want to move it around longer than you need to, further than you need to. So being able to execute close to where you deliver from and be able to get media in and out at a low cost, at a predictable price and efficiently on a global scale I think that's the attraction. So we're seeing big pickup there.
Will Law, Akamai:The other thing that's especially interesting for our users is they have a lot of peak workloads, especially live stream encoding. An event comes three or or four hours, goes away, doesn't happen for a week or a month, and if you have to invest the capex to go buy an encoder that's only being used three hours a week. It's tremendously inefficient, right, but you can come and get a VM and start encoding and you can have it for a day and then you can give it back. So you only pay for what you need and that makes much more sense. So it's the story of Opix, both on the customer side and also on the CDN side. Curiously enough, we focus on keeping our Opix low so that our prices are reasonable, so that our customers can have good Opix. They can see my service.
Dominique Vosters, Scalstrm:Yeah, we see it as well. Also on the origin side, before the CDN, the whole thing, before we see the similar like cost and total cost of ownership. These are these days way more important than 10 years ago. And then this is also, I think, where we come in and work closely together with Akamai, because the platform runs on the Akamai connected cloud and we can run in any edge close to the customer and then from there we can get the signal, transcode it in all the ABR formats and just repackage it in the right format and deliver it to the CDN. And then this can be a 24-7 channel, but also event based channels. And then I this is, I think, ties into in those comments like, yeah, you don't need to have an encoder standing there for a week and then just use it for three hours, just spin it up when, when you need it.
Mark Donnigan, NETINT:Yeah, yeah, yeah, we're, you know, yeah, we're really excited to have dedicated hardware in Akamai ScaleStream. I know, Dominic, you guys take great advantage of that. Arcadian is able to then build services and, for really the first time, the whole promise of the hybrid cloud is actually now, if you really think about it, possible, because everything broke down. You know, as is usually the case, things in theory, you know, look perfect, right, but in actual practice and application it gets more complicated, it gets more complicated, and so the whole idea of being able to quote flex into the cloud is great, as long as you're only using commodity software on CPU, and so then you could have your own data center, maybe flex into the cloud, or sometimes even go the opposite way for certain types of applications. But now, with this edge network, you get that benefit of very, very high, efficient video processing, video encoding. You know hardware that's sitting right on the CDN, that's able to do all kinds of interesting things in real time, and you know that opens up a lot of very, very interesting architectural opportunities.
Mark Donnigan, NETINT:You know, dominic, one of the things I know when you go to the ScaleStream website, you know there's a heavy focus on ease of deployment. I think, yeah, you know, throughout your website, on all the various pages, your marketing, your messaging, you know it's something that you guys really, really emphasize and so I would love to you know have you explain what does that mean? You know, I think everybody says their solutions are easy to deploy, but you know what does that mean for your customers? And then you know how does that impact positively, these trends of these large media companies who need more efficient ways to distribute content, both to reduce cost but, you know, also just be able to get to market faster, faster.
Dominique Vosters, Scalstrm:Yeah, definitely. I think probably on our website you will find three key differentiators. I think the first one is scalability. I think we can do many channels on one server. I think it's one place. We replaced we discussed it before 48 servers, which has two.
Dominique Vosters, Scalstrm:But if it comes to the flexibility, first of all it's the architecture, so it's a fully microservices, so in that sense we can deploy on-prem on the cloud, hybrid or in Kubernetes or whatever. So we have the flexibility there. But the ease of use goes also into the software. So, for example, if I go to, let's say, an example that we use, we have customers that we have together with Akamai. It's an esports company and typically they have events that they need to spin up quickly. They used this similar setup before on a competitive platform and for them it took hours to configure everything, to set everything right and then press the button when the event started.
Dominique Vosters, Scalstrm:When they moved to Akamai and SteelStreet together, they have an API call. They just launched one API call. Everything is set up immediately in the cloud. So we spin up the instance, we set up the channel, it's transcoded in all the APR formats and they're ready to go. And then these esports companies sometimes they have events in one country, two hours later they have an event in another country, and then in this way they can spin it up really easily and this makes it quite efficient. And of course, it's also valid for tier tier one operator. For them, yeah, if it's easy to use, they need less uh people uh on the operation side, or perhaps not less, but uh in a more efficient way, and they can spend uh more time on uh on other things. So I I think it's, it's, it's a really important factor and it ties into the the cost as well.
Mark Donnigan, NETINT:Yeah, interesting.
Will Law, Akamai:Actimite took a bet on NetIn last year because CDN compute vendors typically don't want to deploy custom hardware because it's not portable right and you have to make decisions ahead, a year ahead. You have to buy hardware, install it, say, in Singapore, and then find out, oh, the market demands a Rio de Janeiro, but there's no way you can move the streams to Singapore. That's not cost-effective. So we're always hesitant. But we looked at the media market and we decided that the VPUs offer such a compelling density benefit for encoding that there was no way we could put these in and people would not buy them.
Will Law, Akamai:I think ScaleStream are evidence of that and we have other customers as well.
Will Law, Akamai:So the fact that you can, if you go back, say, four or five years, to IBC or NAB and you wanted to go, look at an 8K encoder it was this 4U device that was warming up somebody's booth. You could sit on it, you could have a campfire up somebody's booth, you could sit on it, you could have a campfire. And now you can go and get a VM and you can, five minutes later, you could start doing two AK encodes on one quadrat on your VPU. That's an incredible differentiator and the other part of it that we're excited about is minutes encoded per watt. The power efficiency of VPUs is not something we've mentioned, but there's the environmental side, which is we can't buy clean power everywhere, so we want to minimize the dirty power we have to buy in data centers. And then it's also our largest operating expense is electricity. So if we can help our customers produce minutes of video by using less electricity, it's lower cost all around and it's an awesome environmental message at the same time.
Mark Donnigan, NETINT:Yeah, yeah, the energy factor is is really important and I think for you know, perhaps for a lot of listeners, you know, primarily we have engineers that are listening to voices of video and you know, or engineering leaders, and to a certain extent it's understandable. Obviously they're conscious of how much energy or power their solutions require, but it's sort of it's it's a couple, you know, ticks away from you know they're not writing the check, it's not directly hitting their budget and in some cases, if they're running on a public cloud or on some you know, someone else's infrastructure, it, you know, there really isn't a direct correlation, they really don't see it, and so it's easy to, you know, I'll use the expression kick the can down the road with power. But I think we, you know we can all agree and I've not met anyone in the industry who would dispute this is that kicking the can down the road is not solving the problem. You know saying it's this public cloud's issue to solve or it's this platform's issue and I'm going to let them deal with it. You know that's not a solution, not only for the environment, which the environmental impact is real.
Mark Donnigan, NETINT:But I was talking to someone yesterday who runs a very large network in Europe, and they're in Amsterdam is one of their major data center locations, and in the Netherlands, I mean, there's just a. You know, I believe it's a moratorium. There are no more data centers that can be built. You know none. So therefore, you need to upgrade your capacity. You need more space. What are you going to do? And you know, in this case, dedicated hardware and just more efficient solutions are the only way, you know, to expand your data center capacity. And you know, I think those are things that you know. Sometimes we're not in the industry always directly aware of, but you know we should be and it's important to note.
Dominique Vosters, Scalstrm:It's definitely on the sustainability side. I think it's all over in RFPs that sustainability and eco-friendly is way more important these days.
Mark Donnigan, NETINT:I agree.
Joe Waltzer, Arcadian:A lot of our clients are asking the same thing, like how do we be more environmentally friendly? How does this impact? All those things that matter to them and matter to their customers?
Mark Donnigan, NETINT:yeah, you know, that's that. That that's interesting, joe, because you're dealing primarily, I believe, with us based, you know, studios, you know kind of hollywood, um, uh, and, and are they operating their own infrastructure or are they going into colos, or are they using clouds, or what does the architecture look like for them?
Joe Waltzer, Arcadian:I find most of them are in the cloud right now. There was a big surge, you know, probably 10, 15 years ago or somewhere around there, where cloud was the big deal because it scaled really well. It was easy to stand things up and we all know the advantages of cloud and I think they all want to stay in the cloud and they want to keep those advantages. I think we'll, you know, said hey, you don't want to buy a, an encoder that's going to use for three hours, you might as well rent somebody's and the cloud will be there for everybody.
Joe Waltzer, Arcadian:A lot of them, as they get more efficient and try to reach again more customers and sort of deploy their infrastructure in more places, are finding that they want to bring it on-prem and a lot of them have minimal infrastructure that they're sort of expanding. We get a lot of requests for how do we sort of repatriate some of this into our data centers, into our stuff? One of the reasons we can do that is because things are just getting more efficient. It's cheaper, it's easier. You don't have to buy really, really expensive hardware to do transcoding. You can buy a fairly cheap box and put a VPU in it and it doesn't cost you that much.
Joe Waltzer, Arcadian:It does a lot of what they're doing now, unless you're dealing with something that's really high demand and really large volume, there isn't a whole lot of reason to do that in the cloud. You can just you can transcode your VOD catalog pretty easily on a couple of PCs in your data center and when they start realizing that, they start asking questions about how do we sort of make that happen. And then the cloud is always their backup, because you know they all think they have a volume and they understand their volumes, but then some big come in and so the cloud will always be there. I don't think anyone thinks the cloud's going away, but we're getting a lot of questions again about efficiency and how do we make this work for us in a way that sort of gets us to where we need to be as far as cost? Yeah, it's, it's one of our biggest questions, for sure is.
Will Law, Akamai:There's a lot of analogies to hybrid cloud. People ask like is it cloud all the time or not? And one is to your own car. So if you think about it, your, your car is super expensive say, $50,000 for a car and, like me, my car is sitting in the garage unused most days and maybe I go on a 20 minute ride. So clearly, having a ride share where a car just arrives, takes me somewhere and drops me off, is much more efficient for me. But if you're a taxi driver and you're driving all day, then ride sharing your car, renting it, is expensive, so that's the use case for owning it. So there's certainly media workflows that Joe just elucidated that should not be in the cloud. They're much more efficient being in your own data center and they're typically going to run constantly. They're a baseline load. But as soon as you have variability or have peakiness or sparse consumption of an asset or a resource, that's where you want to move to cloud-based utility.
Dominique Vosters, Scalstrm:Then I think you can even go more in detail, because we also see use cases that even functionality within the platform are split. For example, if you have live channels, this is quite static. You don't get extra channels every year or every day. So this is typically then deployed on-prem but, for example, like thought streaming or uh for packaging, or a server-side ad insertion, either our services that are way more uh, peak events or uh and so on, and these, these functions are then in these, this case, deployed in a cloud, that you have a kind of hybrid setup. Sometimes it's even within the same platform that functionalities are split between on-prem and cloud, absolutely yeah, we see that a lot.
Joe Waltzer, Arcadian:What can we repatriate? What makes sense to repatriate? What should stay in the cloud? Always slicing little pieces out and trying to figure out where we get the most efficiency. It's becoming very hybrid now Very hybrid the cloud's also used to experiment with new technology.
Will Law, Akamai:Yes, you don't know if you want to spend your money on buying it, but in the cloud you can go rent it. Try it out what's the latest, maybe decide. Oh, that speech-to-text works. This one's terrible. This AI agent is awful, but you can. The cloud gives you immediate access to new tech in a much lower barrier than if you had to fully invest in everything. Try it out, decide you don't like it and then try a seller.
Joe Waltzer, Arcadian:It's great for development too. We use it a lot as sort of a dev place where all of our developers can have their own machines and all this stuff. Because, again, when you're developing that application for the first time in the data center, you can't buy or you don't want to buy a bunch of hardware and you know punch holes in your data center and give all these developers access while they sort of build out these solutions. So we use the cloud a lot for, hey, let's stand up the architecture up here, I'll make it work in the cloud, and then we can point it to sort of more a production data center environment. So the cloud's fantastic for development, for research, for figuring things out. We use it a lot for that.
Mark Donnigan, NETINT:Yeah, amazing. So, joe, you're in an interesting position because I think it's fair to say you know you are largely acting as an integrator. Is that sort of how you would characterize I know, you know? And again, when you say I'm an integrator, that can mean so many different things.
Joe Waltzer, Arcadian:Yeah, for sure, absolutely.
Mark Donnigan, NETINT:So there's a question here behind this, but just to make sure I'm framing it correctly. So you know you are partnering with your customers, with these large studios, to deploy technologies, build platforms, even operate platforms. We went through a time where you know, just out of, and Will you know, you're in from the very beginning of, you know, you've been involved in the industry and building technologies and you know everything had to be built. I mean, there just really weren't that many commercial solutions and the commercial solutions weren't maybe mature enough. So a lot of organizations build on open source projects, you know. And then came kind of the early deployments of the cloud and the SaaS platforms, and I'm thinking even of, like, the OVPs. If we remember those online video platforms, you know, we can all think of some of the OGs who you know.
Mark Donnigan, NETINT:Sadly, some of those companies have come and gone, you know even, but you know so then everybody thought, oh, you know, I don't want to build anything and I certainly don't want to operate it, so I'm going to go to an OVP, and an OVP was was valuable for a season, you know. And then we went through a, you know we went through kind of another migration where then the commercial solutions got more mature and people kind of went wait, you know, I kind of want to choose the best encoders, but then I want to, you know, maybe I want to develop my own packager, but I want to have, you know, this other component that I you know, and and so the point in that story, in this you know preamble, is that you know, we've gone through is a very interesting journey of of how workflows are built, what they consist of, what the legacy is are built, what they consist of, what the legacy is. I'm really curious, you know. The question for you, joe, is where are we today, you know, are we in a build? Are we in a buy? Are we in some, you know, combination?
Mark Donnigan, NETINT:What are people both wanting to do and then actually doing? Because sometimes that can be a different thing too. You know, I can want to buy something, but maybe I have some need. I have to build it, or vice versa.
Joe Waltzer, Arcadian:Yeah, I think it's probably more hybrid than it was before, where most people, when I started, were just sort of licensing technology and then using that to do the distributions and they would buy a platform essentially. And today that's still there. It's certainly low risk, right. Developing stuff is expensive. Building a team to develop stuff isn't very cost efficient, especially at the beginning. There's a lot of upfront before you see any results. So a lot of my clients, when they start, want to start with sort of this white label solution let's go license something, let's build some integration points maybe here and there to get our stuff working with their stuff, and let's just launch the service, because time to market is obviously quicker too. Yeah, and I think that's what a lot of people did for a long time, because the economics made sense, because when you're streaming these big movies and you know people are watching them in the millions, it's easy to make those numbers work. And I think now, as we're seeing distribution sort of spread out, those economics don't quite work and the prices you have to pay to fully license all of your tech and they just don't work for some of these movies from the 80s and 70s and just stuff that's sort of sitting in that catalog. So they're looking at new ways to do it and we're getting this hybrid approach where we're doing a lot of integrations with lots of different partners that provide very specific pieces, and a lot of it is where do you want to put your risk? Do you want to? Where can you tolerate the most risk as far as failures in technology? Right? So if you're doing a live service, you're streaming live stuff. You want to keep that in the cloud. You want to keep that with a known provider. That's a risk because if that goes down in the middle of your production, your service isn't going to have a very good time, right? So we have a lot of discussions about hey, what can we offload on-prem and build for you and integrate with different partners and save you money? And where can we not manage the risk? And where do we have to have proven technology partners that can deliver? We know they can deliver, we don't have to test them out and we can get you to market as quick as we can. So it's hybrid.
Joe Waltzer, Arcadian:Again, I think there's a lot of people just having conversations about what to bring in what makes sense and those are complicated conversations, which is why you know we've had a lot of them lately because but when you hadn't, when you did the ovp, you just called someone there and you could have a couple people and you could do the deal and there you were done. You didn't have to hire a team to make that work. You could hire, you know, a few people to get that done. And now, when they're building their own internal solutions, the teams get larger, the knowledge gets a little bit more technical. You have to build, you know you have to hire a much larger team to make that work. And it works out when you do it in the numbers.
Joe Waltzer, Arcadian:But it's a complicated task. You have a lot of old school people who still like the old way where it was simple and easy. They just had to sign some deals and someone else ran it for them. And now you have this sort of new generation coming up that wants to build it and help out, and they're all sort of coming together and they're trying to figure it all out. It's very different. Everyone does it a little differently. But those are the conversations we're having.
Mark Donnigan, NETINT:Yeah, and Dominic, you know. So it strikes me that you then sort of almost sit a little bit in the middle right of being a managed platform but then also having the software blocks, the modular pieces, the applications right that somebody could deploy. What do these trends look like, you know, for your business and for where you see things are going? You know, are you becoming more of a platform? Are you becoming more of, you know, kind of a modular approach? You know, are you building more bespoke? Because that's a requirement as you engage, you know, with more customers?
Dominique Vosters, Scalstrm:Yeah, no, it's actually a very good question and the answer is not that easy. So I natively I read a software company and so I read a little bit of software. I'm a software company, so I deliver software and let's call it products or solutions in different areas like packaging, transcoding, shield. So that's our core focus. Of course, sometimes we see content owners or smaller customers typically don't have the experience to do everything in-house and then they come to us like can you do a managed service for us? And then it depends a bit on the use case. So if it's a simple setup, then often we do the managed services. But for larger projects we typically rely on partners.
Dominique Vosters, Scalstrm:Because we are a software company, we try to focus on where we are strong at and then we want to keep building on this and then not focus on areas where other companies might be much better than us. So if you say, do I need to draw a line, then we're a software company, we not manage services, sometimes case by case. We might do it, but that's not the default. And then also our products. They kind of evolve a bit along what our customers demand. And then it was winded this road because typically if customers want a managed service. They want an end-to-end solution and then say, if you go back three years in time, we could deliver the origin or the packaging and that's about it.
Dominique Vosters, Scalstrm:We had a lot of products around but we couldn't do the encoding, transcoding and so on. So lately in the last two to three years, we spent a lot of effort on delivering transcoding as well. Initially we thought offline transcoding, then thought just-in-time transcoding, now also live transcoding for 24-7 channels or just-in-time channels, and then now we can deliver kind of an end-to-end chain. Like if you take this esports company, yeah, I think there you can consider that we're the kind of managed service partner, although the customer has a lot of experience and technical knowledge in-house. So also they take a lot of the management of the platform on their end. But yeah, that's a bit case by case. I would say.
Mark Donnigan, NETINT:Yeah, I understand and Will you know, it strikes me that Akamai is infrastructure as a service largely. How are you partnering, then, with the? You know, let's call it the application layer, because I think one of the challenges and I can just even imagine that there's somebody more than one person, I'm sure is listening to this discussion right now and says, wow, absolutely, I would love more efficient ways to encode my video. Wow, absolutely, I would love more efficient ways to encode my video. Wow, it would be amazing to get away from the high egress cost of public cloud and some of the other. You know the cost factors around the traditional public cloud, so I would love to, you know, consider Akamai. But you know I basically have a workflow that is a combination of I'm buying pieces. You know, maybe my other provider, my other platform, is able to bring kind of a more of an end to end or a complete solution. So what do I do, you know? So how is Akamai looking at this in terms of providing more solutions to make it easier for someone?
Will Law, Akamai:who's not If you've followed Akamai back since 1998. And media there was a point, probably in about 2010, where we were trying to build every solution you could possibly need in the media space.
Mark Donnigan, NETINT:I remember those years.
Will Law, Akamai:Solar media they had orange everywhere and we had a big diagram and we had all these blocks and we were intent on filling in all the blocks so you could do satellite ingest, you could do encoding, transcoding, speech, ad insertion, everything Pretty much all the stuff ScaleStream does today, plus more, and had players and we were trying to do it all and we did not succeed in doing every little bit of it. We did part of it and the challenge there is if you do part of it, your customers are still forced to go out and fill the block. They're forced to become what Joe is, which is a systems integrator. They have to put pieces together. So our new approach, I think, is far more sustainable. We want to sell infrastructure, we want to sell a place for you to run compute workloads, and then we partner with companies who are much better at doing the things that we try to do, like ScaleStreet being an example. So we have a whole environment. We have a program of these trusted partners. People are known to work well on our platform, are integrated, who offer support for getting up on our platform, and when you combine these together, yes, you can start now, have several, all the pathways you want from a media workflow that might be an off-the-shelf end-to-end solution in a public cloud. You can now recreate that, but in a far more flexible way.
Will Law, Akamai:I'm also seeing a trend, I think, if you know the CDN space multi-CDN has been a thing for 10 years. It makes a lot of sense. Not every CDN is perfect all the time and as the cost goes down, I get more CDNs, I combine them as a system and I get reliability out of that group, and this is how CDNs work today. I think the same is going to work for cloud infrastructure. People are very much monocloud today and there's a risk that cloud can have a problem. It's not the best all the time everywhere and workloads have been sticky.
Will Law, Akamai:Cloud providers try to hook you with a solution. Maybe it's the database entry, or maybe it's backup or storage or cold storage that only they have. So now you're stuck with them, and then they can raise egress fees and you're still stuck with them. So people are starting to design cloud solutions for portability. Basically, if it's Kubernetes, that's all I want. I will take care of everything myself, and now I have the freedom to both port it for price but, more importantly, duplicate it for reliability, so I can run in three clouds the same code, the same software. They're all well encapsulated and I get higher overall product reliability from that than I do running it in a single cloud. So I see that as a definite trend and that's something Akamai is trying to accommodate. It's very easy to enter our cloud solution. It's very easy to leave it, but we think that is the future of cloud. Yeah.
Mark Donnigan, NETINT:Yeah, yeah, we're really happy to be working with Akamai because one of the very practical hurdles that we ran into and I think every hardware vendor runs into this is so you know, it's relatively easy, actually it's very easy to ship a card. I mean our VPUs are U.2 PCIe. You know you say, hey, what's your office address? Okay, ship it here. You know you ship it. Somebody goes in the lab, they open a machine, they pop it in, they run some, you know they start testing it Right. So we found it was very easy to kind of get through the initial. You know I'd call it the validation phase. Validation, like, does this thing actually produce quality that's even remotely acceptable, you know, which is the very first thing you know, regardless of software, hardware and encoder, like you have to validate it. And then, of course, there's a lot of more, much more stringent tests that follow, but where we ran into just incredible hurdles and with the largest streaming services where you would assume like, oh, surely they have multiple labs, they, you know they have, you know sort of walled gardens or you know like they're they're prepared to do this level of testing. And we found that they weren't. You know they weren't. And so when it came time to be like hey, we like your quality and we'd be like awesome. So now, like let's get into a POC. And they're like we don't know how to do that. Like we don't, we literally don't have a way to do it. You know, and this happened time and time and you know, and after a while you sort of scratch your head and go, wow, okay, so it's really great to have um, a network, um, like akamai, you know where it's sitting, right there. You can easily spin up a vm, like you say will. It is so easy to access. And this will be the extent of sort of my sales pitch, my plug in this particular episode. But really, if anybody's listening and you've just wanted to dabble, you're intrigued. But it's like, okay, I've tested in my lab, but I, I kind of actually want to put some streams like sort of through the network. You know, maybe not fully in production, but it's easy to do it. And I know Akamai, you know is, is is very generous, even when I think a little bit of you know, initial initial help there, but so Will.
Mark Donnigan, NETINT:I am curious about a topic that I know is near and dear to your heart you spend a lot of time talking about media over quick, and you know low latency and live, and there's something that there's a there's a trend that clearly is occurring and that is the old days of sort of. I have, I run a VOD, you know workflow, and then I have my live, you know. So I sort of have these two systems running to me, and so I'm asking a question here are we seeing a convergence where really streaming is just moving to a live, effectively a live topology? Um, what are you? You know, what are you seeing? And then you know, I'll let you then talk more about you know a little bit of your work around. Uh, you know mock and you know low latency and things like that.
Will Law, Akamai:Yeah, media over quick is interesting. Media over quick is the most interesting evolution I've seen in media delivery in the last 15 years. Adaptive segmented media was the big change in 2010. And now we're starting to see leveraging quick. And why is quick different? Quick has parallel streams as many as you want. You can have 100, 1,000. 2010, and now we're starting to see leveraging quick. And why is quick different? Quick as parallel streams as many as you want. You can have a hundred thousand uh and datagrams as your transport layer, and then over that we have a pub sub distribution system that can move payloads, and one of the payloads that's efficient at moving in is media payloads. So we're building a solution and video.
Will Law, Akamai:If you think about it, the encoder is producing video in this perfect order. Right, it's the I frames and P frames and they're coming out, and then we cut them into little blocks. We throw them in a big box. They lose their order. We write a text file to tell you how to recreate the order. We throw that over the wall. The player reads the text file, pulls the objects and tries to resequence them and then sends it to the decoder, whereas media of a quick is font or direct. We just take the bytes as they're coming out of the source and we push them to the subscriber. So it's a natural match to how live media works.
Will Law, Akamai:But economically if you only use it for real-time media, in other words to duplicate the conversation we're having now, which it's very good at, by the way, and you're going to see a lot of demos coming out and I can show them if anyone's interested, at the next trade show. But it's tunable latency because we combine the live with caching. So if you want to play five seconds behind live or 15 seconds behind live, you can. You can retrieve it from the relays cache and if you want to play vod, you can think of it as infinite caching or it's content that is never going to change.
Will Law, Akamai:So vod is just on the continuum of life. But it's right at the end to the point. It looks no different to a live stream. That's finished. I've archived it and now I'm playing it back. So economically, if the same distribution network can serve all those types of content absolute real time, some 200 millisecond, 500 to one and a half second interactive live and then higher live as well as VOD now we can monetize the bytes coming out of that across the infrastructure and it's cheaper for everybody. So I think I'm particularly interested in media over quick and its ecosystem of news streaming formats, new encryption formats, new authentication and token usage formats, because it's all confined I I think it's.
Mark Donnigan, NETINT:It's quite an exciting departure from where we are yeah, um, I I'm happy that you gave that preview because I, um, I know that you know, everywhere you go, you take every opportunity it feels like anyway, maybe not literally, but you to evangelize I was not going to discuss media To evangelize media over quick, and it's something that I guess I'm curious about. As we look at, as architectures are changing and you know systems are being built differently, does it open up? You know so this, I'd be curious to get your comments on this To be careful it's not curing everything.
Will Law, Akamai:People are going to carry on using HLS and Dash Correct. This is running just fine and the end consumer doesn't care what the protocol is.
Will Law, Akamai:They see a block of rectangle and if that works more reliably and at a better price than some other newfangled thing, no company is ever going to adopt it. So mock has to offer a compelling benefit to even make its way into the future and I'll be very clear about that. And I think the benefits will start with real time and they'll expand, but it's not going to be an immediate conversion by any means. Economics is king for all our businesses conversion by any means.
Mark Donnigan, NETINT:Okay yeah.
Will Law, Akamai:Economics is king for all our businesses.
Mark Donnigan, NETINT:That's right, yeah, yeah, yeah, that's right, that's right. Well, this is great. I know we could keep talking. You know, we all collectively bring, you know, so many deep experiences in the industry and you know technology and in the business, but maybe, maybe we can wrap up here, you know, with a little bit of closing thoughts and I think one of the things that just naturally happens.
Mark Donnigan, NETINT:You know, we all attend conferences. You know we listen to panels, we, you know we listen to conferences. You know we listen to panels, we, you know we listen to discussions. You know, like this one, and you know it's easy to listen and say, wow, that's amazing. You know, yeah, I'd really like to take action on that. I'd like to consider, you know, something in my own business, in my own workflow. But then, you know, the question always is, but I'm saddled with all this workflow. But then, you know, the question always is, but I'm saddled with all this legacy and I'm, you know, or I'm, locked into, you know, agreements with vendors or agreements with cloud providers or whatever, and so there can sometimes feel like I really want to, you know, be innovative, but I'm kind of stuck. What sort of um, you know, tips or hints, or, uh, you know, what sort of advice would you give somebody who's listened to this last 45 minutes or so and you know, is saying, wow, that's amazing. But now, how do I? How do I transform? What can I do with?
Dominique Vosters, Scalstrm:this. Any insights? Yeah, I think, start small. If you have an existing platform and you want to test out new stuff, you don't need to throw everything away and start from scratch. Of course, yeah, it's good for our business, but typically it doesn't work like that.
Dominique Vosters, Scalstrm:But you can start small. You can start using or change bits and pieces and, for example, uh, even these channels, uh, move them to a new platform, we move them to the cloud. If it works fine, I it will work fine, and then I join or move other stuff to the cloud or on-prem. Or even you can use existing hardware, because you can gradually move stuff to to other platforms to use the other existing hardware for for another component. So I, if you say I'm stuck with my existing platform or contracts, you can always be innovative and add functionality like I don't know, add insertion or just-in-time packaging, where you say, okay, I want to save storage space, remove all the lower profiles on storage, do the packaging on the fly from these profiles. So I don't think there's an excuse to not innovate. Yeah, I would say.
Joe Waltzer, Arcadian:Dominik for my first response. But my second thing would be would just go ask people, you know, go to a conference and just ask questions, because one of the great thing about this industry is there's a lot of passionate people. I love what I do, I love talking about what I do, I love hearing from other people, industry and having these kinds of conversations where new things come up, where new technologies are discussed, when your opportunities are presented, and I think that maybe you think people won't be very forthcoming, but I think you'll be surprised that some people will just give you the secret sauce and they'll tell you what they're doing and how they do it and why it's special, and maybe they will try to sell you their service. But at the end of the day, I think most people will just say, hey, this is what I think, this is your problem, this is how I approach it. Hey, maybe you should talk to this company. This is their solution. It might be a little better than my solution.
Joe Waltzer, Arcadian:I think the industry is very open to just having great conversations, because everyone's situation is different. Everyone has a different issue, slightly different ways they need to solve their problem and just different realities. I think that just having that conversation. And just go to a conference, walk to a booth that might have a solution that you might need and just go talk to somebody there and just start having a talk might have a solution that you might need. And just go talk to somebody there and just start having a talk. I think you'll you'll learn a lot, you'll find out a lot and you'll you'll start finding some solutions to your problems and having somebody who's what to do next my advice is slightly different.
Will Law, Akamai:I would tell people is sit down and think, because how we make video is changing under our feet, as it is an existential threat to all four companies in this webinar right now, which is if I go to Facebook Reels and I flick through it, I would say today, 50% of the videos I look at were not encoded. They were made by AI. Right, those did not go through any type of encoder. There was no media workflow, there was no production system behind them. It was servers in a server farm from you name it AI vendor making video, and that's already 50% of what I'm consuming. So next year that will be 75% or 80%, very, very quickly.
Will Law, Akamai:Only a few percent of the video, the minutes we watch, are actually encoded by the industries we all work for, and I find that, on one hand, a scary proposition. Right, because that's a threat to our business. On the other hand, it's an opportunity. If that's how video is made, how do we participate in that production? Right now it's short-form content, but that's just a limitation of how fast our computers are. Eventually that will expand to a full movie. You want a custom movie with exactly the people like you like and the setting you want, with the theme and the prop and the story and the language. It'll just be made for you and it won't be made by the products and services we're all selling today. So I certainly don't have an answer to this, but I think it's the question worth addressing and that's what I would encourage people to do today is think of that environment and then think how you win if that's how we make money.
Mark Donnigan, NETINT:Yeah yeah, amazing insights. Well, we live in super exciting times I don't think anybody would debate that and things are changing fast. But a lot of opportunities. So with that, thank you very much, gentlemen. It was a wonderful conversation Really. I enjoyed it. I'm sure our listeners did too.
Joe Waltzer, Arcadian:I had a great time. Thank you, Mark, Thanks Will. Thanks, Dominique. Awesome, Thank you, Mark. Thanks Will.
Mark Donnigan, NETINT:Thanks, dominique. Awesome, and I just want to close with this Again, at IBC, if you are attending IBC, make sure to stop by the VPU ecosystem booth. So it is the NetEnt booth, but a lot of the branding says VPU ecosystem. All of these companies ScaleStream, arcadian, akamai, and there's going to be a few others that are joining us there are going to be there doing demos, talking. It's just going to be a really, really great opportunity. We're running a mini conference in the booth, so we also have sessions and, again, all of the companies represented here today are going to be speaking. So just a tremendous opportunity, not only to learn, you know, about what VPUs can do for you, but also what these companies are doing in general, even outside of VPUs, just in video. So make sure that you look us up if you're in Amsterdam. And with that, we thank you again for listening. Be safe, be well.
Joe Waltzer, Arcadian:Thanks, everybody Appreciate it. Bye, this episode of Voices of.
Dominique Vosters, Scalstrm:Video is brought to you by NetInt Technologies. If you are looking for cutting-edge video encoding solutions, check out NetInt's products at netintcom.
