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
No Silver Bullets: The Messy Reality of Video Streaming Technology
The streaming video ecosystem has come a long way, but as Jason Thibeault candidly puts it, "The streaming video tech stack is a mess." In this illuminating conversation, the Executive Director of the Streaming Video Technology Alliance (SVTA) takes us behind the scenes of streaming video delivery, revealing both the remarkable progress and persistent challenges facing the industry.
What started in 2014 as a small consortium of 17 companies has blossomed into an influential organization with approximately 130 member companies collaborating on critical streaming technologies. Unlike traditional standards bodies, the SVTA focuses on creating immediate, practical solutions through specifications, best practices, and even open-source software that address real-world streaming challenges.
The heart of our discussion centers on what makes streaming fundamentally different from traditional broadcasting. While broadcast television benefits from dedicated infrastructure, streaming navigates the unpredictable terrain of best-effort networks. This fundamental difference explains why even sophisticated platforms occasionally struggle with buffering, jitter, and reliability issues that broadcasters rarely encounter.
Particularly fascinating is Thibeault's breakdown of technologies like open caching, which creates overlay networks enabling heterogeneous caching infrastructures to function as unified delivery systems. He also unpacks Netflix's groundbreaking achievement of streaming to 60 million concurrent viewers during the Tyson-Paul boxing match - a watershed moment that provided valuable lessons for the entire industry about scaling live video delivery.
Looking ahead to 2025, edge computing, security technologies, and improved advertising infrastructure emerge as key focus areas. Meanwhile, the role of artificial intelligence in streaming will likely center on specific applications like content recommendations, personalized clip generation, and predictive operations rather than transformative changes to core infrastructure.
Ready to dive deeper into the technology powering your favorite streaming services? This episode offers invaluable insights whether you're a streaming professional seeking technical knowledge or simply curious about what happens behind the scenes when you press play. Subscribe to Voices of Video for more conversations with experts shaping the future of video technology.
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. Good morning, good afternoon, good evening. Depending on the part of the world that you're in, you are listening to another episode of Voices of Video and I am super excited. I'm here with my friend, jason Tebow, so welcome Jason. Hey, Mark, hey everybody, how's it going? Hey, yeah. So they're all saying, hey, we're doing well. We're here to listen to you, jason, so you know I was reflecting you and I have known each other for a number of years now. Going back to, I think we were saying limelight when you were running strategic marketing and you know I always admired I can remember picking your brain. You guys are just cranking out the best content you know. Technical, it was in depth it. It answered real questions. It wasn't the fluffy marketing stuff. Um, I think the industry has gotten a lot better, um, uh, today, but boy, 15 years ago or 12 years ago, it was pretty sad.
Jason Thibeault:There's still a lot of this revolutionizing, transforming, evolutionary.
Mark Donnigan:Transforming. 18 reasons why I call it the 18 reasons why marketing.
Mark Donnigan:If I see it's an automatic delete you know, and like I, when I see seven reasons or seven factors, it's yeah, yeah, forget it. No, I love it. I love it. You and I are both marketers, but you know we we also are, you know, technical, and so, of course, you took over the streaming video alliance. So, of course, you took over the Streaming Video Alliance, which you'll get to tell the story about, sva to SVTA a number of years ago now. I'll let you tell the story, but, seriously, welcome to the show and we're going to have a great discussion. So, yeah, why don't we start there? You know, for those that don't know who you are and are saying streaming video tech, what? Who are these people?
Jason Thibeault:Yeah, no, no, I mean, you know, the SVA at the time was a streaming video line, started in 2014 with, I want to say, 17 member companies, and the impetus was really, you know, alon Mayor from Quill. You know he, I think he got this idea of like, hey, we were in this transition of transparent caching and you know, but we knew operator networks needed caches at the edge. So he, I think, kind of came up with this brainstorm idea of like why can't we democratize caching, make it an open standard? And so he went and sat on people's doorsteps and knocked on doors and got people interested. And, you know, the SVA was created and so originally the impetus was around sort of a place to develop this idea of standardized, democratized caching infrastructure. But obviously there were other working groups. Even at the time there was, you know, some security stuff and some geo stuff and some other networking and transport stuff. It wasn't just open caching, but largely it was open caching. That was the focus, yeah.
Jason Thibeault:And so they realized a couple of years in that they you know that in order for the SBA to really grow, they needed to get somebody neutral in, and so at the time, limelight was a board member and Joe DiPaolo at the time was on the board. He's like, hey, I know this guy at Limelight, he. And Joe DiPaolo at the time was on the board. He's like, hey, I know this guy at Limelight, he would be perfect. And from there to here, that's what I've been doing. I came in to lead the organization. We added a T to the name a couple of years ago in order to make people aware that we were not touching on business issues. We weren't talking about monetization, we were talking about business models. We were really addressing streaming tech. And we're talking about monetization. We're talking about business models yeah, really addressing streaming tech. And obviously the organization has grown considerably. I think we're now close to around 130 member companies.
Mark Donnigan:Is it.
Jason Thibeault:Yeah, most of the streaming who's who you know, the companies that really have an impact in the industry, are participating. Just this year alone, we published 23 documents and 23 new projects started. Wow, we added two new working groups this just this past month one for encoding and packaging, so, you know, to capture best practices, because we have all this stuff like context where encoding and it's great for specific use cases, but everyone's kind of you know doing it their own way, and so it's like, why don't we share this information with everybody? Yeah, that's what we are right. So we are a what would normally be called a standards definitions organization, an SDO. We're very much like IETF or CTA, wave or W3C or SCDI or SMPTE. We have working groups and everybody meets and they decide on what to work on and then they create some deliverable whether that's a document, a spec, the best practices or even software code and then we give it away, right, um, but we don't do standards, and the reason for that is the streaming video industry is super moving fast.
Mark Donnigan:Um, standards take and very political by the Very politically correct statement there.
Jason Thibeault:Jason, yeah, so it's just moving very quickly and by the time you standardize something, people will be like we're not doing it that way anymore. We figured out a better way and so we stick with things that people can use right now, like either specs or guidelines or best practices or, again, software code, um, and, and the organization you know is really really expanded. You know, our mission has grown a bit. We now have a university, um, you know, we have public repo where you don't have to be a member to participate, um, again, all these projects. We just launched our first industry group for streaming sports called the streaming sports council.
Jason Thibeault:Uh, and that's, and that's to capture the technical challenges. It's not to solve them. It's just to say, like, what are the technical challenges for delivering, you know, high quality live sports events at scale? And you know we'll feed our working groups with those challenges to to potentially solve them with, with solutions. Or we'll feed other organizations, we'll feed industry events like conferences, like SportsPro and SVG, you know, so that they have technical topics that are germane to the sporting industry, rather than just trying to guess. It's like no, no, no, these were surfaced by people actually in the industry.
Mark Donnigan:Doing the work.
Jason Thibeault:Yes, exactly, and so that's where we're at now. I mean it's you know, again, we're, we're, we're continuing to grow, continuing to try to do things. This year we merged with DashIF, so anybody in the industry probably knows that there's a lot of associations and if you want to participate in them, you know you're paying membership fees for different ones and it's just, it's a pain, it's a pain. And so Dashjs recognized like we really want this work to continue. We're at a point where you know we're continuing to iterate on the Dash spec, but we're also maintaining the Dashjs player base and you know some other things, and it's like we need to get someplace where this integrates well.
Jason Thibeault:And you know there was either us or DVB. They were talking to both of us and they selected SVTA because we were a little less bureaucratic. You know we wanted to kind of let them do what they were doing, and so this year it's been great we have a DashIF working group. Now. It's kind of the same as it was as a separate organization, but now you just join the SVTA and you get to access both the SVTA work and the DashIF work and that's worked out really well and we'll integrate a little bit better next year.
Jason Thibeault:There's some duplicative functions underneath DashIF that would probably be better served by being part of a larger SVTA working group. But that's the story of the SVTA, where I said we're doing really, really well. It's a fantastic year from a growth and membership perspective and especially projects and work output. It's just been amazing. All those volunteers. They're absolutely fantastic.
Mark Donnigan:Amazing. Yeah, that's great, thank you. Thank you for that. Now, do you have to be a member to be able to access all this wonderful documentation, or is it freely published on the website? It's freely published.
Jason Thibeault:Yep, that's really sort of. What we're trying to do is trying to improve interoperability in the industry, and you can't do it by, you know, gatekeeping your content you can't put it on a paywall.
Jason Thibeault:So everything that gets ratified by the membership, so it goes to a ratification process, we publish openly on the website um, and then again everything that's not ratified yet, that is members only. So members get to work on drafts and uh, again, like our software, we have a a pretty large repo base um that the members are working on, but there's one that's a public repo. That's the common media library, which is an attempt to solve some of the player code-based fragmentation, and we allow anybody to come and participate.
Jason Thibeault:They just have to join the repo, agree to the DCO and then there they go, it's managed by somebody at Paramount Casey, and so people say like, hey, I really want to be a committer, I want to play that role. Then they just reach out to Casey and you know they work it out. And so it's a really, really cool project that we were allowed or enabled to make happen this past year as a public repo, which is super exciting for me. And we've got lots more coming. So there's 2025 is going to be a lot of software.
Mark Donnigan:Yeah, amazing. You know, and I just want to make sure that, for those who weren't that familiar with the mission of the SVTA and what exactly you guys do, you know, I'm going to paraphrase, so if I get it wrong or if you want to put an exclamation mark behind anything, feel free to jump in. But what is so special and unique and, I think, valuable, you know, in my view, to the SVTA compared to all of the other different bodies, and every group performs its own functions. So you know, it's not that you know, but what's very special is that the industry right now, I would argue always is clamoring for best practices.
Jason Thibeault:Dare I even say SOPs, which goes way back.
Mark Donnigan:But I mean, this is you know, jason, you get to work with all these 130 plus member companies. You've been in the industry a long time, you know, I've been around a long time and you know, um, yes, there's amazing people doing work in the largest streaming platform companies in the world, all the way down to the smallest. But you know, even if you're in the largest and you've got this massive team and you and and everybody's a true expert in their respective areas, you can't know everything. You know and and and I'm, I'm always you know.
Mark Donnigan:I don't say surprise because it just it stands to reason that you know I'll engage with somebody from some major, well-known platform and find out, wow, they're looking for answers in some cases, just like you know, a much smaller company is you know and and and again, not because, oh, you know these people don't? You know, don't quote, unquote. It drives me nuts when I read these comments on LinkedIn, that sort of insinuate the company X or company Y doesn't know what they're doing, you know it's like. Or company Y doesn't know what they're doing, you know it's like. Anyway, we could go down a whole nother rabbit trail on that one, but I think the SVTA is a incredibly valuable resource that more of the industry should avail themselves to. So you know, that's my-.
Jason Thibeault:No, absolutely I mean, we sit in the middle of the industry, right? We cover the streaming workflow, which is basically, once content hits origin, everything from origin to playback and analytics. That's all what we cover. And you know we are a forum for collaboration. Our working groups are all bottoms up. They decide on the projects they want to work on. So anybody can come in and raise their hand and say, hey, I've got this problem. We saw this and you'd be surprised. 10 people in the room would be like, oh yeah, my God, we had that problem too. Hey, should we try to solve this with some best practices?
Mark Donnigan:or some guidelines.
Jason Thibeault:We even launched a streaming video operations group which is focused on operations. So Mark Ison at ITV, he was like hey, can I get a place where we can just talk about operations and operating streaming platforms and experiences with vendor technology and can we improve? So yes, what we want to be is a place in the middle where people can come and work together with their colleagues across companies, come and work together with their colleagues across companies, so even alongside competitors, to try to solve the interoperability issues that make the streaming tech stack kind of messy. It's not like TV, it's not a broadcast, it's not based on standards, it's hard to hot swap stuff when you want to switch tech, and so here's a chance to come to an organization and work with people to try to fix that problem.
Mark Donnigan:Yeah, yeah, amazing, amazing, well, great. Well, the title of today's discussion is exploring the present and the future of streaming technology. So, jason, I know you brought your crystal ball for the future, but before we crack out the crystal ball, we bring out the crystal ball. Let's talk about the present. And we've got. By the way, I want to put a nice shout out to everyone who registered in advance for this session and who submitted questions.
Mark Donnigan:We got some really good questions and we got a few more than we usually receive. So thank you very much. If anybody is watching live right now, you can go into whichever platform you're on and submit questions and we'll do our very best to get to it. So get to them. But you know, jason, why don't we just jump in and I'll let you kind of pick where we are today. You know there's all kinds of trends. What comes to mind when I think about this is you can't, you know, trip into the front door of NAB or IBC, or you know.
Jason Thibeault:Mile.
Mark Donnigan:High Video, or you know, I mean you know whatever Name the conference and name the trade show, and not see the words plastered everywhere Ultra, low latency, low latency, mock, webrtc, Generative AI, yeah, I mean, yeah, and the list goes on and on and on and on. Fast and all this stuff. Right, and you know, we're laughing and sort of poking fun, not because any of those you know technologies or technology references or capability references are funny and not useful and valuable, but I think we also can acknowledge that there's a lot of words being plastered on walls and not too much following behind. Yeah, so you know what is the state of all those things? Oh my, yeah, yeah, all right, there we go, and you have 11 minutes to answer that question, so go.
Jason Thibeault:You know I'll continue to beat the drum that the streaming video tech stack's a mess. It's gotten better, don't get me wrong. So there are definitely and have been improvements, but it relies on technology that's not necessarily just in the industry, right? So because it's software-based, the tech stack is software-based. There's lots of changes within the way that we develop web-based software and middleware and things like that that are impacting how we're building software for the tech stack. So it's messy. We're trying to fix that.
Jason Thibeault:So when we look at technologies like Mock and or Quick so, for example, our networking transport group has been doing lab testing of comparing Quick versus TCP for video delivery under different circumstances and in different network conditions. So different network control algorithms, different compression algorithms and you know quick is not the silver bullet Like fancy that there's no silver bullet. We always seem to be looking for one but there's not. So when we look at the state of things today, I think the true drivers of priorities are things like reliability, resilience, scalability oh, wait a minute, it's the same things that we talk about with broadcast television. Yeah, that's because I remember this one thing back from my days at Limelight BBC was. You know, we're talking to BBC, and BBC was like listen, we don't care about all the bells and whistles, but it just has to work so somebody can sit down, press the play button, like they do today on the couch, and press the remote control on button. Just has to play and has to play flawlessly, and let's get rid of the buffering and the jittering and all of this other stuff.
Jason Thibeault:But obviously, streaming is different. It's going's going over a best effort network, that's right. It's not a self-contained network, and so there's lots of issues that you have to contend with. You don't have to contend with when you're literally delivering signal, and that kind of sums up where we are today. Right, we are struggling and trying to figure out how do we make this scalable, and so that involves things like edge resources. But you know, how do you host edge resources in the best possible manner for things like live streaming? We just spun up an edge group to start asking and dealing with those questions. There are so many elements and components of the delivery chain that have not been addressed yet.
Jason Thibeault:And then we have to look at things like the state of content delivery networks. How are they going to fit in this ecosystem that is starting to see more operator networks like Comcast and Verizon and Orange and Telefonica and all these people launching their own CDNs in order to better deliver to their end users? And gosh, that makes a whole lot of sense. You know, if I'm Disney, why do I want a relationship with a bunch of third party networks when I can just go straight?
Mark Donnigan:to the iBone networks?
Jason Thibeault:Yeah yeah, but we're seeing some of the technology like open caching kind of shape around these conversations that are happening at the business side and that's and that's the funny thing, all these words that we talk about like fast, and all this stuff. You know the media loves to jump on words that reflect how they think people are watching or experiencing something different, um, but they don't pay attention to the technology underneath. Fast is nothing new. Fast is zero new. Exactly, we were doing pseudo live, which is what? Fast?
Mark Donnigan:is we were doing pseudo live 15 years ago.
Jason Thibeault:Um, yeah, it's not, it's not anything truly new and it's just a different way to package stuff. Um. And so we see all these words and it's just like is that? Is that really reflective of what's happening underneath, which is where all the hard work is being done? All of those smart engineers and the smart software developers and architects all across these amazing companies, from, you know, from the big guys like Netflix and Apple and Amazon and Google, all the way down to the startups, and Amazon and Google, all the way down to the startups, you know they're trying to fix or build the better technology.
Jason Thibeault:underneath all of those cool words, and I want to talk about mock for a second in particular. Please do.
Mark Donnigan:Please do.
Jason Thibeault:So you know that's a weird story, right? So Google, you know, comes up with this idea of quick based on HTTP3, let's use this. And then they're like, well, you know what, this would probably be best if we stuck this in the industry. And so they give it to IETF and it becomes IETF Quick, because they start working on it. But Google's like it's not happening fast enough, like we're using this and we need these features. And so then it became Google Quick and they have since aligned, which is great. There's just one Quick now. But then Quick was not really designed to deliver video, and so along comes this idea of media over Quick. So how do we tune, optimize and build features in Quick that are specific to media, like my favorite feature? There is TCP fallback. What happens when video goes to the enterprise and they're blocking udp whoops people to watch a video? So let's have a fallback mechanism that you can configure for a tcp based stream.
Jason Thibeault:Um. So mock is mock is a good thing, right it's. It's a it's an improvement. But quickock are really like. They're kind of like the next generation of WebRTC. It's like the WebRTC that WebRTC should have been, because WebRTC is never designed for video, exactly.
Jason Thibeault:Yeah, so again. So all these things under the hood that we're trying to do as an industry whether it's Mock or it's context-aware encoding or it's AI what we're trying to really do is ensure a seamless, reliable, resilient, scalable infrastructure. That's what all of this work is intended to do. And there's other work people are doing on viewer experiences and multi-camera and you know all these cool features and AR and immersive and XR and that's all great, but that all relies on the stuff under the hood. So we are I still consider us to be in a building phase. We are, um, there are services that are running obviously commercial services, and in a lot of ways, they're duct tape and bubble gum, and I say that, as they are basically will do anything they need to do to make the service operate because they're generating revenue. Yeah, that's right. Not be the best way to do it from a resiliency, reliability scalability standpoint.
Jason Thibeault:But you got to start somewhere and you got to make it work, and so my hat's off to all of these streaming platforms that are making it work. But then our growth and our membership and the projects we're doing and the amount of work that we're doing shows that they all realized there might be a better way to do there might be a standardized way to do this. Let's all get together and make that.
Mark Donnigan:Yeah, yeah, I really love that perspective, jason, because and one of the questions that someone submitted is around Netflix, and you know how they're going to do better on Christmas day you know they have, they have, they have, they have two games on Christmas day, right, nfl games and and you know, hey, I mean you know we don't have any insights and if we did, we couldn't say it anyway. So you know, so you know so there's lots of speculation, right.
Mark Donnigan:There's a ton of speculation, but you know, as I've said very publicly, you know, shortly after the Tyson-Jake Paul fight, you know, like, look, here's the one thing that I do know and that we can all take to the bank, there's a group of incredibly smart people who are incredibly hardworking at Netflix who are doing everything possible to make sure that that experience is not replicated. In other words, you know, the less than stellar experience is not replicated on Christmas Day that we can take to the bank. But the thing that I the reason I reference, you know Netflix and this question the person is I love your perspective is that you know these large platforms have grown. Like you say, there's a little bit of duct tape, sometimes a lot of duct tape, under the hood, and you know bailing wire and you know whatever, and it's not a bad thing, it's not, and it's yeah, it's not a bad thing and it's also.
Mark Donnigan:We have to step back and look at you know we are now 15 years into a lot of these platforms. You know it's hard to actually like, like, like, really think about it. Like, a lot of these platforms came out of the ground in 2008, 2009, 2010, 2011. That makes them like 15 years old, which is kind of hard. It was just kind of hard to believe. And then you think about the state of you know how the system was built then and how it evolved over time. And when you're growing like hockey stick and you know, and you're, you're just running to just keep up. You know, even if, even if at multiple junctures throughout that period, you know the engineering team says you know, gang, you know, like, like, like, this system needs to be fundamentally rebuild or re-architected or whatever. Like you, you literally cannot do it.
Jason Thibeault:It you know you can't build the plane in flight.
Jason Thibeault:Yeah, I mean rebuild the plane in flight you know, and that is that is a great way to to, that's a great perspective, and the reason for that is still to go back to netflix, right? So netflix launched uh, an event that they had no idea was even possible to host and deliver, which was a 60 million concurrent event, blew every record we've ever had completely out of the water. They didn't build the Netflix architecture with that in mind at the beginning, because they were delivering, bob.
Mark Donnigan:Yeah, exactly, and just remember, Jason, it's only like three or four years ago, and I won't say the senior member of the team's name, although you know I remember pinning him down oh, this would have been 2015, 2016, maybe and saying you know, is Netflix going to do live? You know everybody, netflix needs to do live. And I mean he looked at me with so much certainty just like Reed Hastings when he said Netflix will never have ads and looked at me and said Netflix will never do live. We are VOD, we're we're happy with that business model. It ain't going to happen. I mean, he didn't say the part, but you know it was like it was that definitive of an answer.
Jason Thibeault:And now hello and and guess what?
Mark Donnigan:he wasn't wrong at that time. At that time, that a hundred percent was the correct and the accurate answer. You know, our businesses are growing, they're dynamic, they're changing and our technical needs then you know, and that's the funny thing, right.
Jason Thibeault:So again. So people ask the question you know, what is netflix doing better? That's the funny thing, right. So again. So people ask the question you know, what is Netflix doing better for Christmas game day? The question is like what did they learn from that experience that we, as an industry, don't know and they learned a lot.
Jason Thibeault:They learned again delivering a live event to 60 million people and it's funny. So we've been talking to NASA. So NASA is going to stream live from the moon in 2027. Hello, like what is that? That's a global event.
Jason Thibeault:That's not behind a paywall uh, can you say a hundred million people? Yeah, yeah, really like. Yeah, you know, netflix is cheap, so netflix is such a pioneer they are. They have taught us so much about streaming at scale. Yeah, and they are teaching us even now. They're they. They basically said, hey, we're going to take a chance on this event. I have no idea what it's going to be like. It's going to be 30 million, maybe maybe 20 million, people watch it. Ah, there's 60 million and there's so much in the way like there's isp interconnects, there's teleport facilities and transits and isps and home bandwidth all this stuff that has no bearing on delivering a signal to a cable, to a television set, and so they've learned so much. I'm actually really eager, because they've been very good in their tech blogs, to come out and talk about things that hopefully someday they will talk about like hey, here's what happened and here's what we learned and maybe that'll be.
Jason Thibeault:Be honestly, maybe that'll be post-NFL Christmas Day. If Christmas Day goes down a hitch, they'll be like ah, we learned, implemented, we optimized, look, we did so much better. Now let's talk about what happened and what we changed, and so that's again. That's what the SVTA is about sharing information. We cannot get better as an industry unless we share, and Netflix has been so good about sharing. Um, they, and actually they they recently signed, uh, you know, their membership agreement, so they're very eager. Oh, amazing, awesome, awesome. The svta, yeah, so it's especially when we have groups like encoding and packaging, where they have a real passion, um, and are really looking to improve. You know how they deliver content from an encoding perspective. So, yeah, so I'm I'm very excited and and very appreciative of companies like Netflix who are willing to be honest they're willing to take the risk for the rest of us to improve and do better.
Mark Donnigan:Yeah, yeah, for sure, it's great, okay, well, if anyone wonders where Jason and I stand on this whole you know discussion debate, you can see. Yeah, yeah, yeah, it's pretty clear. So this is a great discussion. Now another question that came in and it dawned on me. You talked about open caching. I would not assume that everyone knows what that is. So the question that came in is what's the state of open caching? So why don't you first give an explanation, a definition, talk about open caching, and then what's the state? Where is it being used, adopted, what's happening?
Jason Thibeault:Yeah, the best way to look at open caching. It's an overlay network, so the idea is caches can be open caching enabled. So it's a suite of api specifications. It's not a new reverse proxy. It's not a new nginx or varnish. It's a way to enable existing caches to be viewed as a single footprint. So, for example, like let's say that comcast cdn was open caching enabled, so so they've implemented the specs in their existing caches so that the APIs can be used to communicate with them, so upstream and downstream CDN communication. Let's say that Egeo did that. Let's say that somebody else, some other company, company, fastly did whatever you know. Um, you've got places that are doing caches and and the content owners? Right, we all know the streaming operators have caches. All right, they all have. They all have their own content delivery networks in order to shield origins.
Jason Thibeault:It's kind of like a virgin shield exactly all of those caches are open caching enabled, then the content operator, when they're delivering content, can see an entire footprint and control the delivery of content through those open caching enabled caches, which would mean things like they could purge simultaneously from multiple delivery networks. So that's what really open caching is all about. It's all about creating an overlay network on a heterogeneous delivery architecture, and so it's. You know, the biggest probably provider of open caching right now is Quilt. You know and it's funny I always say that it's you know, quilt, it's like the Red Hat model. Right, you can go get the community version. So that would be, you go to SVTA, you get the specs you build out what you want to build out.
Jason Thibeault:You build it yourself. You build it yourself and your support is only the community. Or you could do enterprise Red Hat, where you're paying a fee and having somebody manage it for you and there's some special extra secret sauce features that are not available in the community version. So same for NGINX, same for Varnish. That's secret sauce features that are not available in the community version. So same for Nginx, same for Varnish. That's kind of what Quilt is doing.
Jason Thibeault:So, quilt, you know operator networks. You know operators are saying we need caches at the edge. That's done. We need caches. Saves us on middle mile backhaul. It's a problem caused by Unicast and we could talk about that forever. But they need caches.
Jason Thibeault:And so operator never saying like, hey, we'd like a cache that could serve all sorts of content.
Jason Thibeault:It's not specific to a type of content or owned by a specific content operator, um, and and so quilt is coming in and saying like, hey, we got a box, you slide it in, we manage it for you. You know, we help you. To be honest, the big thing is about traffic federation is we help you share some of the revenue. So you, you, you get your your little piece of that delivery revenue you know as well, versus uh cdn, which may terminate on an operator network and the operator is responsible at no revenue generation for backhauling that content through their network and down to their subscriber. So um, so open caching is really about democratized caching. It's an overlay network that simplifies and organizes and makes it easier to manage heterogeneous delivery networks and it's all API-driven, so you don't have to build something new you can like. Hopefully in the future we'll have open caching plugins for things like Nginx and Varnish and you can just make a little plugin yeah and you get it.
Mark Donnigan:Yep, you get it for things like Nginx and Varnish, and you can just make a little plugin and add your caching and you get it. Yep, you get it. What's the corollary to open caching? And what Netflix built with OpenConnect?
Jason Thibeault:So there's no correlation, right? So OpenConnect is Netflix's proprietary CDN caching box that they put into specific operator networks it and and it's obvious why, right, it's obvious why an operator would want this box. Uh, netflix accounts for like 30 of internet traffic during peak times.
Mark Donnigan:Yeah, I want that content cash.
Jason Thibeault:I don't want to backhaul it, yeah, and so you know, I don't know where this goes in the future. I mean, what they've built with OpenConnect is really fantastic. They've really built a great box.
Mark Donnigan:And they've got a massive footprint now. I mean you know the numbers that are out there. I've gotten a little bit of exposure to maybe what the real numbers are and people would just be I jaw dropped well, I mean again, look at, sort of look at, look at akamai's footprint, their caching footprint.
Jason Thibeault:They also embed in operator networks, which is the obviously good strategy, yeah, and so they have a massive footprint as well and that helps operator network. So operators have a you know, they have this like thing to weigh in their hands is do I, you know, I can't only host so many boxes, right? So power, yeah for the premium. Um, you know, do I, do I host an open caching box?
Jason Thibeault:or two, or do I, you know, can I, can I host that and akamai and open connect, but then disney comes along with their own exactly next thing.
Mark Donnigan:You know, my racks are like 50 60 percent full of other other people's, you know this is very much in flux.
Jason Thibeault:Open caching is again. It's not a silver bullet um, it is a a potential part of an overall delivery strategy yeah, but it relies on a lot of other components and parts and participants that you know, uh, a Netflix doesn't have to worry about and Akamai doesn't have to worry about, like, and so there there's, there's trade-offs and and that's why you know you're not getting this everybody rushing to adopt and implement open caching, because this is not an easy decision for an operator to make.
Jason Thibeault:Um, and they're you know, they're going through that process again. Quilt provides them an easy way to get that which is, oh, I can just put in quilt boxes. This is fantastic. Um, but again, it's still a trade-off. Do I have the rack space for a lot? Do I want to manage it? Do I want to have the power, like I already got akamai caches in here and I've got open connect caches in here and maybe've got OpenConnect caches in here, and maybe that's good enough. Yeah, this is probably another 10 or 15 year journey.
Mark Donnigan:Yeah, yeah, ok. Well, that's the answer to the question of the status of open caching. And you know, again, you said a very important phrase. You know, open caching is a part of an overall strategy. Important phrase, you know, open caching is a part of an overall strategy. And you know there's going to be a couple themes that come out of our discussion today. I know, and you know, one of them is that there's just no silver bullets. And so you know, with all due respect to the vendors that we love and I know I can think of some vendors that I really like and I greatly respect but they've got to stop talking as if you know, what they have is a silver bullet, you know and and so yeah, so talk about that again.
Jason Thibeault:I want to. I want to bring up akamai again. I mean, they're doing a lot of great stuff, like they have a managed cdn product, which is an operator network, and say I want a cdn. I don't build myself, i't imagine myself. I could just go to Akamai and get this and that solves a whole lot of my problems. There are so many business decisions that affect the technology direction that we pick, and so that's why they're you know, and and and, so it's just there are just you know again, there are no silver bullets.
Mark Donnigan:Yeah, yeah, exactly, ok, let's transition. We, you know, we've got 15 minutes or so to talk about the future and and, and I know that, right as we were, as we were going live, I think you were pulling up maybe a post from like a year ago. I was, yeah, like, like, like literally a year ago, you know, almost to the day or the week, anyway, you know. So how did you do? How did you do with your 24 prediction?
Jason Thibeault:around for a while, right, and a tech radar divides things up by you know sort of these quadrants of techniques, tools, platforms, and then sort of languages and framework, and then everything's in a radial circle and it's you either adopt it, trial it, assess it or hold it. And so I'm still kind of reviewing what I looked at. But the way I sort of approached my predictions back in 2023, for 2024 were around these big technical things, these big technology buckets, so things like I looked at you know what's the state of cloud, what's going to be the state of cloud in general, or what's going to be the state of the edge, or what's going to be the state of serverless functions, or AR, xr, vr. You know where are we there Ultra low latency streaming, and you know, in some ways I'm just looking at this sort of you know where are we there, ultra low latency streaming, and you know, in some ways I'm just looking at this sort of. You know, half-heartedly, I was pretty close on a lot of these.
Mark Donnigan:Yeah, that's good.
Jason Thibeault:Because I was just really talking about are they going up, are they coming down, are they staying the same? And I have new ones to add, and so I'm going to update my predictions right cool and again can I call them predictions sort of like evaluation of the technology and industry for 2025 and we'll do another tech radar. We'll adjust the tech radar, but we'll go up and down on some stuff and we'll add some new things like server guided ad insertion. Oh, that's a new one, yeah yeah exactly, and but it, but it's.
Jason Thibeault:Yeah, I was pretty close on a lot of these and you know some may still be like we didn't really go down. We're still climbing a little bit. People were still talking about it, but but yeah, not too bad.
Mark Donnigan:Yeah, cool, cool, cool. Well, what stands out? You know what? What? What stands out? You know, let's look at twenty five and then you know, not too far in the future, but you know where? What is on your radar, radar tech, what is on your radar tech, radar, no pun intended for this next year, for this upcoming year. And what do you think we should be paying attention to? Investing in, maybe planning to invest in, technology-wise?
Jason Thibeault:Yeah, I would say, from a technology-wise perspective, a lot of the emphasis is going to be on edge. So that's again is how do you scale effectively, um? And then the other emphasis is really going to be on security. We saw that explode in 2024 holy smokes, um. And then you have, uh, you know, the cta wave. Svta common access token standard was published in December, so now we're going to see, potentially, people starting to implement things like that. I think there's going to be a continued investment in trying to reduce device fragmentation. It's a.
Jason Thibeault:It's a really, really, really big problem like it's a gnarly problem if you talk to these content owners, like if you talk to a hulu or talk to disney, you know, talk to probably even netflix. They have warehouses of televisions with different versions, models, os's, because they you have to physically test these things, because those devices phones, tvs they go out to different countries. As we deprecate them as US consumers, those devices are sent to emerging countries or emerging nations and developing nations and third world countries and that becomes their primary device. And so then you're like OK, if we don't allow that anymore, then we've just cut off content and entire countries.
Mark Donnigan:in some cases you know entire markets.
Jason Thibeault:In some ways that has to be done. There are some cases where the hardware itself, it cannot be secured well enough, so it's like we can't deliver to that box anymore. But yeah, so I think there's going to be a lot of push in that direction. It's funny I had this sort a lot of push in that direction. It's funny. I had this sort of bucket of thing called interactivity and I put it up or down and I might say it's going to be down again. I think we're all the as an industry. So this is this other part of the industry, right? So we have the more technology which allows us to stream reliably with resiliency and scale and then you've got the other part, which is what we'll just call sort of the user experience.
Jason Thibeault:How can I? Because? Because it's no longer signal, it's now digital. What can I do with these digital bits that?
Mark Donnigan:I couldn't do before a multi-camera angle.
Jason Thibeault:I can have um, I can, I can um include in other alternative people talking about, like a sporting event. You know a guy from TikTok, he's got his own channel and he's you know, I can just click on him. So there are some ways where we're trying to shape the user experience to be more interesting, especially to different generations. But interactivity as a whole, shoppable commerce I would have a negative 25. I don't think we're ready for this. I don't necessarily think people are super interested yet, in the future maybe, but it's still not.
Mark Donnigan:I'm aware of one particular platform that is operating primarily in Southeast Asia. It has incredible scale, delivering a lot of simultaneous live streams and doing very well. So it's so also. You know, I remind myself and the folks working around me. You know, it seems like every day, like just because you know this trend or this use case is not hot in the US or in North America or in Europe, or you know some part of the world does not a invalidate it somewhere else in the world or mean that the usage we're hearing about, you know, is sort of like yeah, it's this, you know, sort of corner edge case, like yeah, a corner edge case with like tens of millions and millions and hundreds of millions of consumers that guess what Love to watch these influencers live.
Mark Donnigan:Basically, basically, it's a QVC channel. I mean, that's what that's what's funny to me, it's like we had this when I was a kid.
Jason Thibeault:I watched this on tv and that's the thing is, is there's, you know, there's a differentiator there where, um, you know, a netflix video could be watched if it wasn't just specific to the netflix player, like they didn't, if they didn't have a walled garden. They said, like here's our video, you can watch it anywhere, you can literally watch it anywhere. Tiktok does not exist. Tiktok exists within the TikTok platform, for example, and so it's a very specific, closed community. Now, do the things that they do with their live video, does that maybe at some point and someday carry out? Yeah, so the caveat is, you know, in different geographic regions that may happen faster, may not happen at all, and so, yeah, just you know as a whole sort of interactivity for me is is what I'm seeing across streaming video as a whole is not what we had expected it to be.
Jason Thibeault:I think people are still in the lean back experience. They're just little clips. They're still like I just want to watch, like yeah, I want, like you know and, and we talk about second. How many times we talk about second screen like, oh, people have this app and companion after we all do all this stuff, and are people really doing that? No, they're just they're just on tiktok while they're watching yeah, exactly no.
Mark Donnigan:No. Here's a funny thing is absolutely everyone is doing second screen. It's called phone in your hand, you know, navigating, doing whatever, checking your email not really, you know exactly with with, you know, netflix going or whatever you know know, whatever streaming service of choice. But so, yes, we're all doing second screen. It's just that this whole thing of oh, it's going to be connected and it's this extension of the experience and you can learn all about the actor on the screen, like now, no, but no, you know, 2025, you know, and again, again.
Jason Thibeault:So use cases right 2025, and again so, use cases right, Interactivity is probably far more relevant for live sports AR, VR, XR probably more relevant for live sports or live events. So some of these interactivity things will just be very germane to specific use cases. But again so, I think security in 2025 is going to be a lot of emphasis going to be placed on security technology, and then I, a lot of emphasis is going to be placed on security technology, Um, and then I think a lot of emphasis is going to be placed on advertising, Um. So we're, you know, we're still trying to add tech stack. We talk about the video tech stack being, you know, kind of messy. The ad tech stack is, uh, I don't even I don't have words to describe catastrophe Like it's so complex and it's so.
Mark Donnigan:Fractured, fractured and fragmented, and yeah, it's well okay, so so. So. So, jason, this leads into a couple, a couple of questions that came in what's going to be the hottest topic at NAB 2025. So, in in four months, and or yeah, uh, and what is the role of ai?
Jason Thibeault:in the no, no, you, just, you, just. What's going to be the hottest topic at nab? That's exactly.
Mark Donnigan:That's exactly why I asked those two together. They came in from two different. Maybe it's the same person, I don't really know, but they're two separate questions.
Jason Thibeault:But you know, let's you know, let's get it clear. So I mean, I've been studying AI for decades, you know back in. So if we go back to the history of AI, you know we start talking about folks like Minsky and Lamatt and MIT and sort of what they were trying to do, and I actually I was sort of been penning a blog post about renaming artificial intelligence to autonomous intelligence, because that's really what we want. We want like, right now, the chatbot doesn't exist unless I engage with it, it's not doing its own thing and coming up with stuff and having I don't open the chatbot and it doesn't say to me hey, jt, how's it going man?
Mark Donnigan:hey, I saw this movie the other day it's not an actual conversation, it's not autonomous it doesn't operate on AI.
Jason Thibeault:For me, as we see it now, it's really just clever programming. Large language models and, even more appropriately, small language models are fantastic. So using these to create and using them against natural language processing to create ways that we can interact fantastic. You know AI, so generative AI. So, like looking forward as to how that might impact the streaming video industry. It will like.
Jason Thibeault:I really believe it has a place for things like content recommendations let's, let's improve content. It has a place for personalizing clips that you want to like, like, let's say, for example, for a sporting event, um, or sporting events across the league. Like I just want the recaps of everything, like what would interest me. Yeah, I can take care of that. You don't have, you don't have a manual person sitting in front of going like, let's clip this part and clip this part and clip this part, um.
Jason Thibeault:So ai is going to be hot at mbb yeah, 100 um, and I think that will probably be one of the big stories. Uh, and, and it does have a role in the streaming video workplace. I think it's more has a role on the edge, um, I think, from an operational perspective, ai the way we have ai built now can really do well at predicting where things are going to go wrong and saying like, hey, based on all the data I can analyze in seconds that you as a human can analyze in an hour, I can tell you this ISP is probably going to have an issue because we've been seeing these key metrics go down. You should deal with it right now. That's where I think AI is really going to be applicable is in operations.
Mark Donnigan:Yeah, yeah, totally agree, and you know, as you're talking about this, and I'm thinking about my own perspectives and learnings, I think we need to do a show, an episode, and just focus on only the AI conversation, because we need a whole lot more time, you know, and I'd actually love to bring in, you know, a little bit of a panel of experts, you know, and so anyway, so that's a public pronouncement that we will put that show together and not too far.
Jason Thibeault:We often run before we can walk.
Mark Donnigan:Yeah, exactly, we have to fight right now without really understanding, like, how to use it.
Jason Thibeault:well, and that's where we're at. We're in the sort of like let's throw darts at the board and figure out what sticks, Kind of like. Oh wait, there was another one called blockchain.
Mark Donnigan:Yeah, that's right, let's just apply it to places and see if it works.
Jason Thibeault:It's like why don't we be more thoughtful? Yeah? Yeah, and in terms of technology before we put it into practice.
Mark Donnigan:For sure for sure, yeah, for sure. And in terms of technology, before we put it into practice, for sure for sure, yeah. So I think we'll wrap up here. This question is actually interesting, maybe not applicable to all of our listeners, but I think people care about it. The question is which technologies have the most transformative potential? That, right, there could be a 30-minute conversation, but the second part of it, I think, is where the meat is, and I'd love for you to comment on this one.
Mark Donnigan:The second part of the question is how do we make sure that patent trolls don't prevent their dissemination? So, in other words, there's a lot of great technologies being developed, but then, you know, and codecs are the easiest one, right, like you know, av1 was, you know, was going to be, you know, kind of our savior in that regard. And, for the record, and everybody I think knows from reading my LinkedIn, I'm a massive AV1 fan and very involved in AOM, et cetera. So, but you know, and then all of a sudden it's like, oh, maybe there's going to be some, you know, some, ip questions around, even AV1. So what's your view on this? Not the AV1, but just the patent issues, you know.
Jason Thibeault:Yeah, I mean obviously AI, you know, could be really transformative to a lot of parts of the workflow, from encoding to analytics, you know. So they're sure people have already patented a bunch of stuff. Thankfully there can be a lot of prior art out there, so maybe some of those patents get invalidated and that's, that's great. Like no, no patents is probably probably better for our industry than having patents. But so we live in a world.
Jason Thibeault:We're going to have to deal with the trolls. Hopefully, hopefully, you know companies will recognize the benefit of having these patents out in the open and they'll just make them publicly accessible.
Mark Donnigan:Here we're not, you know you know, like what Elon did with Tesla, right All the Tesla patents. I mean that's pretty amazing. And look, it's worked out okay for Tesla.
Jason Thibeault:Oh, I like this last question. It just popped up about.
Mark Donnigan:Web3. Yeah, so why don't you answer that? Squeeze that in.
Jason Thibeault:I'm a fan, so you know I do believe that D-Pen and Web3 have a place in media delivery. I think it's really, really interesting where this is going. So D-Pin, if you don't know, is about basically decentralizing key infrastructure power, internet, bandwidth, cellular modems why not content delivery? So how do we improve content delivery? By decentralizing it. I think that's where Web3 could really come in.
Jason Thibeault:And somebody you know we have a startup, a grant member in our organization called BlockCast, and I've been helping them out a little bit. You know they are proposing to do just that. It's like how do we get, you know, how do we get caches in people's homes? Like, how can we get caches in people's homes? How can we help those caches be part of a larger caching pool that other people can deliver content from? So it's really early, it's really interesting and I think we're going to see some real transformation happening in the way certain aspects of the streaming workflow are handled, whether it's delivery or analytics or encoding. You know there are some deep end companies out there doing, you know, doing encoding or transcoding as a decentralized service.
Jason Thibeault:So, it's very, very, very interesting sort of what's been what's happening with Web3 and deep end and what we might see in the future.
Mark Donnigan:Yeah, very cool. Well, jason, 100% we're going to have you back and do a follow-up because you know, really we could have just kept rolling and I think everybody would stay fully engaged, you know.
Jason Thibeault:A lot of stuff to talk about.
Mark Donnigan:Yeah, we do, we do, but let's wrap it up here. Thank you for coming on. Really appreciate your perspectives, my pleasure.
Jason Thibeault:And I want to say thanks to all my member companies. Right yeah, comcast, all these member companies are the ones doing all the work and they are coming to the table and they're collaborating with each other and we're trying to build um a scalable, resilient, reliable streaming infrastructure, and that can't be done if everyone's doing their own thing in their own little corner and just protecting themselves. So I just want to shout out to all of my member companies you know I so thankful that they're participating and what they're helping to accomplish in this industry, and it's going to be more to come, so lots more to come, as we keep pushing forward.
Mark Donnigan:Amazing, amazing. Well, keep up the good work and for anybody who hasn't been to the SVTA website recently, make sure you go easy to find in google. Just try, you know. Type in streaming video technology alliance or is it yeah?
Jason Thibeault:svtacom right or it's in the north.
Mark Donnigan:Yeah, okay, cool, sounds good. All right, well, uh, that wraps up this episode of voices of video. Thank you for everyone who joined us.
