Video: The Future of Cloud Transcoding
Tim: Welcome back to Streaming Media East 2017. I'm Tim Siglin, contributing editor with Streaming Media Magazine, Media Strategy Principal with Reel Solver Inc. And I have with me today David Trescot, from Hybrik. And David, who were you with prior to joining Hybrik?
David: Before Hybrik, I was CEO of a company called Rosette and we did product called Carbon Coder which became really the dominant on-prem transcoder for professional use. We sold that company to Harmonic and worked for them for several years and then left, did a lot of traveling around the world and then the whole team got back together and now we do Hybrik.
Tim: Very good, you know, in fact we were talking about Rosette Carbon Coder and some of these other on-prem products the other day. I've been in the industry now for 19 years and I remember sort of seeing all the on-prem. As I understand, you're cloud-based now, so tell me ...
David: Everything is in the cloud now, so what we've done is we've taken transcoding to the cloud but also more than transcoding taken media processing workflow. So in our minds it's not just transcoding, it's also the quality assurance and media analysis and file delivery and all that. All that is moving to the cloud. For really almost everybody is either they've got a cloud plan or they're making a cloud plan. And so for us, we look at this as the right time to kind of be in the market with a media processing system.
Tim: Interesting. So, would it be similar to something like Vantage from Telestream but in the cloud where there are multiple modules of the workflow process?
David: Well, what we do is we kind of compress everything into one system, so it's not something where you have to buy ten individual things from us. Everything is included, it's all built in, and it's also set at a price point where, for example ... One of our modules is in a thousand node configuration. You're not really going to get that from other people. So we do a scale that is pretty much untouched by anyone else.
Tim: You mentioned pricing, what is sort of the pricing paradigm?
David: Well, we do it based on the total number of machines that you're gonna run. In fact, the processing itself you pay Amazon directly for.
Tim: So the instances for processing…
David: Exactly, we're giving you that at the rock bottom price, right. We're giving you that, we help you take advantage of the spot market, so you can get things at the lowest possible price. To give you an idea, I can get a thousand machines on Amazon for 50 bucks an hour.
David: People haven't really gotten their minds around the cloud, what it can do.
Tim: What types of instances, though, 'cause obviously with Amazon you've got instances that are micro all the way up to ...
David: These are XLarge machines, C4 XLarge machines ...
Tim: So you're saying a thousand XLarges for ... how much an hour?
David: These are four core machines for 50 dollars an hour for a thousand machines on the spot market. Now, what that means is that we think the traditional companies in coding space have looked at the cloud as just a "Oh, let's do our exact same business model in the cloud that we did on-prem."
Tim: Right, as opposed to changing the business model to fit the cloud.
David: We took everything from scratch and said, "What would you build from scratch in the cloud?" with that idea of "How do you take advantage of a thousand machines?" And so the things that we're doing, for example, we've developed all our own software so we don't have to pay licensing on a per machine basis so we can spit up a lot of machines. And back to your pricing question, for us, a thousand dollars a month gives you ten machines, $5,000/month gives you a hundred machines, ten thousand dollars a month gives you a thousand machines. So really, when you compare that to anyone else, people are offering cloud instances they might offer pricing of $2,000/month per machine ... Ours at our highest pricing is $10/month per machine. So we are literally 20 times cheaper than other available services.
Tim: What about something like Linode or Rackspace who also do cloud at $5 a machine, I mean are those ... Are you beholden to Amazon or is it something that you're actually going to use other ...
David: We actually designed our system to be able to run on-prem or in the cloud. We're not dependent upon Amazon, but the reality is Amazon is 90% of the available cloud market. They have the best machines, they have the lowest pricing, they have the spot market. So we find ourselves and our customers ... I mean, we'll go where our customers are, because transcoding needs to happen where the data is. Really, it's a storage problem. Even if Rackspace had free transcoding, free machines, it would still not make sense to necessarily move your data from Amazon to that and then back.
So really, the decision that our customers have to make first is where are they going to be? Where is their data? And once you make that decision, well then it follows on that the transcoding happens in the same place.
Tim: Okay, and then one final question: you mention that your system was designed for on-prem or in the cloud, you obviously thought through the cloud model and how to approach that ... Where is the sweet spot for somebody who wants to do on-prem? Is it a security or is it price point?
David: We actually have no customers that we're doing on-prem with. All our customers are cloud. We believe that it's clear that that's the future. That's where people need to be going, that's where people need to make their investment. If you don't have a cloud plan, you're behind the curve. The cost structure, again ... Media processing is inherently a spiky workflow. And trying to deal with a spiky workflow with an on-prem solution is almost impossible. And you can't do the things that would've made sense like, we've introduced Netflix VMAF, the Video Multimethod Assessment Fusion. One of the great things you can do with that is you can encode multiple times to find out what the correct value should be. So you encode, you see whether "Is my quality too high? I'll lower my data rate." On-prem you never do that because you can never afford the time to re-encode it five times. But if five more machines cost me ten cents to do an encode to check, then yes, I can do that in the cloud. So workflows become possible in the cloud that simply are not possible on-prem.
It's also, as you go into 4k, HEVC, Dolby Vision, HDR10, HDR10+, all these things essentially preclude on-prem processing. It's not going to work; you have to be dealing with the cloud.
Tim: Okay, all right, very good. David, thank you for your time.
David: My pleasure.
Companies and Suppliers Mentioned