Behind Discovery’s Move To IP And The Cloud

Since joining Discovery Inc. from Fox Cable Networks 2003, Chief Technology Officer John Honeycutt has led the construction of Discovery’s technology center in Sterling, Va., while overseeing the company’s expansion into international markets and its embrace of IP-based technologies.

He has integrated the technical operations of European cable sports giant Eurosport, which Discovery acquired in 2015, and is now busy with the Scripps Networks properties that Discovery acquired in March.

His pace isn’t likely to slow down anytime soon, as Discovery announced Monday a 12-year deal with the PGA Tour for its non-U.S. broadcast rights. To exploit the right, Discovery plans to launch a PGA-branded OTT streaming service.

TVNewsCheck Contributing Editor Glen Dickson recently spoke with Honeycutt about Discovery’s technology path based on industrial supply-chain thinking and the power of the cloud.

An edited transcript:

You began your career on the sports side of the TV business, working with the regional sports networks at Fox. What did you learn there that has had the biggest impact on your career at Discovery?

Obviously with our Eurosport business there is a direct correlation because Eurosport is essentially in the regional sports business except it’s done on a country-by-country basis.

And if there is ever a business that needed that, it was that regional sports business. There was analog satellite at that point, we were just going digital, and you could spend a lot of money pretty quickly. You had to be really cognizant of geography and blackouts and commercial insertion, and it was — and still is — a really complicated business.

But looking wider, I think that I learned about the intersection of technology strategy, business strategy and execution. That’s sort of a broad statement, but my point is that it takes all three.                                                                                                                                                   

When I came to Discovery, it was a complicated business because of the global footprint, the different regulatory requirements and the different customs or consumer preferences within markets.

Learning to manage all of that early on in my career was just super, super helpful, and I think I learned mostly how to be in way over my head.

I was young. I think I became an SVP at 27. When you are standing there and you have got David Hill, Jeff Shell and Peter Liguori and Tracy Dolgin and Peter Chernin, Chase [Carey], and those guys are looking at you to execute, that was a pressure-filled situation. I really learned how to just swim in the deep end of the pool from that, and that’s just been super helpful my entire career.

One of the things you have talked about for a while is this notion of a supply chain for broadcast operations, and you were talking about that long before a lot of other CTOs in the broadcast space. Can you describe what that means in the media world, and how Discovery has implemented it so far?

You have to look at our business and just think about the key tenets of what we do. You know, we either produce an asset, so we are in manufacturing there, or we acquire an asset, so we are going to market and try to procure product.

Then you are obviously having to manipulate it and package it and then quality-control it. You are having to customize it, and ultimately you have to ship it and put it at the point of consumer. All of those principles align to a supply-chain mindset.

Unfortunately, for us in the media business, we never make exactly the same asset twice, but the concepts are all the same. I am getting material from one of, in our case, 600 production suppliers from around the world. So, I have to have standards in which I accept material from them. I have to time the acceptance of those materials so the product reaches the consumer at the given appropriate moments; whether that’s from programming, or whatever the demand from the business is to get it there.

And then increasingly we are delivering assets to multiple distributors. It used to be that it was a linear television model and you would essentially go to market in one place. You have got to have one grocery store, so to speak or one style of grocery store, which was a linear television business.

But today, you have multiple [stores]. You have your own OTT services, you have other packagers, meaning Amazon, Netflix, whoever they could be. You have VOD for your suppliers.

So, you’re having to think about distribution on many, many different levels, and if you are not managing the expectation of when those assets are going to be available and what you need to do to them and then the timeframes to do that, in a business as big as ours, you can get in a lot of trouble in terms of not being able to deliver the assets quickly.

If you can’t measure it, you can’t manage it, and we measure a lot. We measure throughput; we measure mean time to execute, how long does it take us; we measure capacity, in terms of resources available to us; we measure utilization, because now we have the opportunity in the cloud world to be able to decrease costs at times when we’re not using resources.

It used to be you built for the max. If I knew that I was going to have this many hours of transcoding to do, I would go buy ten transcoders. And then you look at utilization of that, and it’s incredibly low across a 24-hour-day, seven-day-a-week view of the world. But with the cloud now, I can actually tear that transcode engine down, for example, and not pay for it.

So, it really makes a manufacturing mindset come to the forefront when you see all of these capabilities. And we have embraced it. We have embraced it hard at Discovery, because we see having to supply a global footprint across multiple platforms. And if you don’t have that kind of manufacturing mindset, it’s very difficult.

That said, we are in a creative business. So, you have to give producers and creators time, and if you have good command of your supply chain, you can maximize the time in which the creative people have the ability to make assets.

This all can’t be simple manufacturing in the end. This is an art, the business we live in, but if we apply some of these principles to components of it, we can give the artist more time to create the best possible product.

Discovery is seen as a leader in adopting cloud-based technologies, particularly with the move of the U.S. channels to cloud playout last fall. Can you talk about the company’s migration to the cloud?  When did you start thinking about using the cloud, what was your first application and where are you today? How much more can you do?

Why don’t we talk about where we are now, and then we will go backward. So, we moved all of the U.S. playout into the cloud last fall. We are currently in the process of migrating what we call our European prerecorded channels. So, think non-sports. Put Eurosport aside, along with our Scandanavian, free-to-air channels that have a lot of sports in them. We haven’t migrated those yet; that’s coming. Where we are today is all of the U.S., and we are maybe a third of the way done on the European channels. We will be done by the end of the summer, and that represents over 150 channels for the European business alone. So, it’s big business that we are moving. 

Where it started [in early 2015] is looking at the supply chain. We said, where are we here in the process and what are our friction points, and what don’t we like about how we are handling assets and then ultimately distributing assets? The playout part gets a lot of attention, but the bigger move for us was actually the front end of the supply chain and receipt of assets from production suppliers.

So, we have a project internally we call On Ramp, and basically what that is is the acceptance of assets from production suppliers into the cloud, in this case, to an Amazon [AWS] bucket. What we did is we extended our business systems outward, and this comes back to the supply chain point. It used to be that literally, when I started at Discovery, you would walk in every morning and everybody would be kind of waiting for the FedEx guy to show up, because the tape was coming. You are just sitting there watching, and it’s “Oh, he is late again!” or “What am I going to get today?”

And then suddenly you would be in a situation where you have 32 things, and then you spike on capability because you only built for 30. It was just this very backwards mindset to me. So, what we did is we took our business systems and pushed them outward, and said we know that on Thursday we are expecting this asset to be delivered to us, and we told the cloud that basically. So, it was almost like an expected delivery, and then you could line up resources sitting behind that.

When the asset arrives, I know I have to QA [perform quality assurance] it, I know I have to maybe make an international version of it and I have to prepare it for languaging — all of those supply chain steps that used to be human handoff to human handoff.  We automate it through the process and there are gates in there. If it fails QA, it kicks into a different bucket and a human has to go take a look at it. But the human is told why they have to go look at it because the automation has provided clues as to what the issue with the asset was. 

When it came to playout, we had an aging platform. We were going to need to replace the platform, the automation system, and I was loath to put in traditional on-premises hardware, blinking light infrastructure. I just don’t think that’s the modern way of doing things.

Look at what can happen in software. I will use Netflix as the example. That user experience is incredible, and that’s all software. So, what are we doing that’s any different than that? Somebody is hitting play, you are inserting ads, you are putting the graphic over the top, it’s all the same concepts.

We just needed to do it in a scaled way that was reliable 24 hours a day, seven days a week. And that was hard. That was picking the right vendor, and that was software development to be able to replace all those little components in the hardware chain, little one-rack-unit devices, be it an audio limiter or a Nielsen encoder. We had to make those basically, in software, to do it.

Don’t get me wrong. It was hard to do, but we had a map to work from to make that happen. And we have been successful. And the same thing is kind of happening now. I am seeing a lot more companies who are looking at it saying, you know, this is the way to do it. It is becoming, I think at least in the United States, it’s becoming the industry norm among the larger scale players, that this where we are all going to go, being in a primarily software-based model.

It makes a whole lot of sense because of the flexibility of it, especially as businesses change their size and scope. You either acquire channels, as we have done with Scripps, or you are selling channels, or you are changing the portfolio and maybe shrinking the portfolio in some cases. The flexibility that software brings you is hugely powerful versus a whole bunch of hardware infrastructure.

Just to clarify, when you are playing out, is the signal coming out of the Amazon cloud or is a final feed going to you and then out via satellite?

We are still there [with a satellite feed] for a couple of reasons. One, no one has yet built, much to my chagrin, a cloud-based kind of statistical multiplex environment. You still need that aggregation of the transport stream, basically. And then, secondarily, in the U.S., you still have a lot of smaller operators that aren’t set up to take it directly from the cloud; they still rely on satellite; they are just not capable of having the kind of bandwidth that’s required to go and take those streams from the cloud.

But all of the big guys — Comcast, DirecTV — are fiber. So, the next step in all of this, which will come I predict in the next 12 months, is we say to them: “Just meet me at the edge,” or “Here, I will put it in AWS West. You guys go get it from there.” And that will complete the circle.

In the U.S. that will happen, and I think that it could potentially happen with some of the smaller countries as well. In Scandinavia, it’s already all fiber anyway. There is no satellite. So that’s kind of ready to go. But then you get to Chile, you get to India, where there are thousands and thousands of headends, satellite will be around for a long time. But I think that you will see in specific markets and specific scenarios that we will move to an all cloud-based distribution probably in the next 12 to 24 months at the outside.

Discovery is a huge programmer, and you’ve been talking about distributing to huge MPVDs with massive nationwide infrastructures like Comcast. So how big do you need to be for the cloud to make sense as a programmer?

Well, I am not sure size is gating. I think it’s more about looking at the whole supply chain, end to end. You know, picking certain workflows or certain components and kind of bouncing up to the cloud and down doesn’t make sense, because you pay on egress from the cloud.

So if you have a workflow where you upload as asset to the cloud, distribute it some place, then bring it down to edit, then put it back up and then take it back down for languaging, that makes no sense and you will just get killed economically.

And then the other thing is, you have to look deeper. We looked at real estate costs, we looked at power costs, we looked at air conditioning and whether you are able to shutter a facility based upon our move. [Editor’s note: Discovery is in the process of shutting its London playout hub.]

If you can, all that cost goes away, property taxes go away, all of that stuff that you have to pay for. So, no matter your size, you have to look at everything. And I think you can, even for a smaller programmer, it can make a lot of financial sense if you go all in. If you don’t go all in, you are going to struggle to make the economics work.

Would it be viable for a small call-letter station to do what you guys are doing?

I don’t know. I think certainly distribution of assets from the networks to the call-letter station, that should all be cloud-based at this point. News is still hard. But having a server sitting on-premises that you need to replace and have a maintenance contract on, all that stuff, I struggle to see how that’s going to be viable going forward for anyone, be it a small call-letter station or a big broadcaster.

I think it’s obviously going to be situationally dependent, but I can see very specific things that would be advantageous. And if they aggregate enough of those things together, you could probably make a case.

What percentage of your overall operations still relies on traditional broadcast equipment, and where do you see that being in five years?

For the U.S. business, the only traditional broadcast equipment we have is encoding and uplink. We are in the process of moving, like I said before, Europe to that exact model, and that will be done outside of sports by the end of August.  We will then do Latin America, and then Asia will come after that. So definitely within two years all of our linear playout will be 100% cloud-based, outside of sports. 

The uplink situation will depend on when those markets are ready. But in the U.K., it’s about five locations I have to get it to. So, just take the stream out of the cloud, and then what Sky or those other guys do with it is up to them. But I think in two years, all of the prerecorded linear business as well as all of the distribution for syndication — Amazon, our own OTT business, all of that — will be cloud. 

In sports, we are working hard. We have a very active project that we are looking at with Eurosport and the cloud. We want Tokyo 2020 [Olympics] to be heavily remote-produced. We want to send less people to Tokyo than we had to send for Pyeonchang, and we think the cloud can be a big part of it.

The big issue with cloud in sports is latency. My competitor on that is Twitter, frankly, because if you are watching a game and you have got your phone and you have Twitter open and you see somebody you follow scream “Goal!” and then it takes 20 seconds for you to see it on television, that’s an incredibly disappointing consumer experience that we don’t want to have.

So it’s definitely an issue that needs to be addressed and I think it’s being worked actively. We talk to Amazon about it all the time. We talked to Google about it, we talked to all the cloud service providers about it, that we need to move the latency needle, so to speak, more towards real-time.  It will never be real-time just because of the processing nature of it, but what is the acceptable realm? Is that five seconds, is that seven seconds? I don’t know. We will have to measure and see what we can get.

With broadcast operations moving more and more to IP-based systems, security is obviously a big concern. What makes you confident a cloud-based system won’t get hacked?

It’s just guaranteed that Amazon employs more security engineers than I could ever hope to. Look, their whole business is creating a secure, reliable environment. I trust it. You trust it every day with your credit card, with your consumption and ordering things. It’s inherent in their business. If they didn’t have a secure business, they would be out of business.

We obviously had that conversation with them about their strategy, and we were satisfied. We look at it and we audit it and we talk to them about it. If somebody wanted to get one of my assets, for the last 20 years they could hack into my environment and get it, too.

But we look at it and say that we are very comfortable with the tactics we use — not only our internal information security team, which I think is world-class in the media industry, but we also use external consultants to help us ensure that those environments are secure. If you look at the vast majority of breaches that happen, they are all through phishing and/or social engineering of employees to gain an active password. What security comes down to is, the first line of defense and frankly the last line of defense, is your employees. 

Going to more of an “inside broadcast” question, how quickly do you see the industry moving to 4K UHD distribution? I know Eurosport has done some 4K production. Do you see broad adoption of that, and do you see a difference in how that will be adopted between the U.S. and your international markets?

Well, you didn’t mention HDR [high dynamic range]. So, I believe that this has to be a multiple ingredient situation, and I am very consistent on this point. If you go back and you look at going from SD to HD, there were like five things that happened. Resolution, obviously one. Also, the size of the television. In the United States at least, we went from 4×3 to 16×9. The physical television itself went from this thing that sat in the corner to this thing that hung on the wall as a piece of art. We introduced surround sound at that point, don’t forget. So, there were multiple things that got the consumer to say, “I want that,” not just one thing. 

And I think it’s the same case in 4K. UHD, HDR and frame rate are the three things. So if we can move to a better resolution with a better color gamut, with a smoother visual experience, that works for me as a consumer. I would invest in that.

To the credit of our industry, a lot of us do HD really, really well.

I’ve heard a lot of people in the U.S. have said that if you have 1080p HDR, that’s all you need.

Yeah, absolutely. It’s more bandwidth efficient. You don’t have to throw out a whole bunch of infrastructure. I don’t need to go and create different distribution paths to do it. It’s certainly an interesting option.

I know as a company you don’t want to get into the nitty-gritty details of how you’re integrating Scripps Networks, but can you describe at a high level what your technology strategy is?

Well, we will use scale. That is the answer, to be honest. If you look at their international business and our international business, for example, we have a much larger-scale business. You are starting to see that play out, where we are able to use our scale to pull the SNI business in, in that example.

If you just look at their business, they are fantastic at metadata. If you think about all the recipes they have, their granular ability to manage their assets is really impressive. And we need that in our company. That is something that we are good at, but we are not great at.

I would consider them to be great at it. So it truly becomes, when you go up and down the line, a best-of opportunity. Because they are a really good business. They wouldn’t be around for 30 years and they wouldn’t have built brands as powerful as Home & Garden and Food [Network], and on and on, if they weren’t a really well-run operation. But we see the opportunity to just use our scale to take that even further and help it widen out, from the international example, or aggressively move it to the cloud, which is something that they were starting to do.

They were good in security, we have got great security, so how do we bring those two things together. I mean there is a lot of opportunity for it to be a best-of situation, and to me that’s kind of how it’s playing out.