Solving Common Challenges of Streaming Cloud Migration
One of the common pitfalls of streaming cloud migration involves moving to cloud encoders when streaming to multiple endpoints. LiveX’s Corey Behnke, Cerberus Tech’s Chris Clarke, and TV2 Denmark’s Loke Dupont discuss how to address and solve this and other issues that arise in the course of lift-and-shift migration when transitioning streaming workflows from on-prem to cloud in this clip from May’s Streaming Media Connect.
Encoding via the Cloud
LiveX Co-Founder and Lead Producer Corey Behnke highlights the cost-effectiveness and efficiency of using cloud services for encoding events, eliminating the need for multiple hardware encoders and reducing bandwidth usage. For example, if LiveX is encoding content on X, Facebook, and YouTube, the easy first step would be “to take that one encode, send it into the cloud, and be able to distribute it. And it would save our clients a lot of money because we wouldn’t have to deploy hardware on the ground.” He would usually deploy one encoder per every three streams using tools such as Elemental, Makito, or Teradek. He lists the benefits of this process: 1) “it allowed a remote encoding engineer to run those distribution channels,” and 2) “we had to use less bandwidth for every platform that we had to go to from on the ground.”
Behnke shares an example from last year’s Democratic National Convention—“[W]e had 157 endpoints. Nobody wants to bring 50 encoders to an event, right? The cloud is a perfect place to start with that.” He asks the panelists to identify some issues they’ve seen clients experience when they decide to move from on-prem to cloud.
Distribution and Automation
Cerberus Tech Co-Founder and CRO Chris Clarke advises that cloud migration necessitates thinking differently, noting, “[O]ne of our customers had exactly that example. They would have numerous encoders per endpoint or one encoder per three endpoints, say, so that they weren’t overloading one encoder and being able to separate that off into cloud-based distribution models. But we took that a step further, so actually using an onsite encoder with video and perhaps eight different audio pairs so you get all the different language translations.” Then that gets delivered to the cloud and split off into its own discrete channel that can go to different destinations using a transporting engine. “[Y]ou can take that as it’s coming in and you can repackage it. So you might have some broadcast distribution points that still want an interlaced feed. You might have more bitrate progressives,” Clarke says, acknowledging that there are a variety of reasons for transporting for different profiles, such as how Facebook Live differs from YouTube Live, which both differ from Instagram Live.
All of this can be managed in the cloud, but it’s key to automate the provisioning and the orchestration, Clarke warns. “You don’t want one person responsible in that example for starting 157 different distribution schedules. You would like a system to play that out at the right time so that the operator is dealing with exceptions and issues rather than the status quo,” he explains. He suggests building toward how you would like your processes set up in an ideal world—and chances are, the vendor community is already working on solutions; “you just need to go and scratch the surface and find them.”
Beware the Lift-and-Shift
TV2 Denmark Staff Software Engineer Loke Dupont shares, “For us [it] was [that] we had this problem with smart TVs, and we needed to do a separate encode for that, but we know it was only a limited time, so we didn’t want to go and buy hardware.” The solution was to take the existing encode from the cloud, do a line code for the necessary number of months, and kill it afterward to eliminate the cost, he says. “We didn’t have to go and invest in [hardware] to do that. So definitely that’s a very, very valid use case.”
Dupont sees the lift-and-shift—taking “a physical server and move it into a virtual instance in the cloud”— as a common problem. It leads to compatibility issues and infrastructure instability. If you’re configuring statically because that’s what you do with on-prem, that’s not the cloud approach, he notes. “When you move to a cloud environment, I think the analog that’s sometimes used by software deliveries is that the servers in the cloud [are] not pets, they’re cattle—they’re interchangeable, and you’ll get new ones and then others will get removed. And a lot of software vendors don’t necessarily deal well with that scenario,” Dupont says.
But the cloud is “extremely resilient to failure,” he notes. “You can basically kill off part of your infrastructure and things will continue working. But it does require that the vendors understand that that is something that they need to engineer into the solutions. And sometimes what I’ve seen when people get started with this [is], they do that very simple lift-and-shift migration, and then things end up breaking more than they thought, because they thought the cloud was more stable. And it is. But for it to be stable, it requires you to attune to the way it works.” When something no longer functions, you can “automatically spin up a replacement,” which isn’t always possible for an on-prem data center, he notes.
Understand the Nature of the Cloud
Clarke sums up, “There’s definite pros and cons for all of it.” When it comes to lift-and-shift, “you could take an encoder that sat in a data center for five years performing exactly the same function and never changing. It’s not changed its base operating system, it’s not changed any firmware, it’s not changed any licenses, and it continues to carry out the job. And if you lift-and-shift that into the cloud, within six to nine months, [for] the underlying infrastructure, there’ll be [for example] a Linux support issue.” Next could be the need for new firmware, which requires a new license agreement with the vendor, he adds. “And now you’re tied into a support contract for three years because the underlying infrastructure changed. And I think if you’re not aware that that’s going to happen, it can definitely trip you up.”
It’s important to understand the dynamic nature of cloud infrastructure to fully leverage its resilience, Clarke concludes. “[T]hose benefits are not repeatable in the old way of deploying in fixed infrastructure in bare metal. So there’s definite benefit.”
Join conference chair Andy Beach and other streaming media experts in person Oct. 6–8 in Santa Monica, CA, for more thought leadership, actionable insights, and lively debate at Streaming Media 2025. Registration is open!
Related Articles
Streaming Media presented its 17th Connect virtual conference May 20-22, featuring speakers from Paramount, Meta, Google, NBA, Roku, Warner Bros. Discovery, AWS, Akamai, Fubo, DAZN, Fremantle, IAB, Philo, Estrella, Eluvio, nanocosmos, Cerberus, and more, and session topics ranging from and sports on FAST to biddable advertising to SGAI to AI in live streaming to MoQ and more.
27 May 2025
COVID accelerated the push to IP-based streaming workflows for many companies, proving that live streaming at scale via the cloud was possible, although not always in tip-top form. Live streaming has been something we've been covering since this publication started, but here we are looking at content that has traditionally been delivered via production trucks, satellites, and a lot of on-prem resources.
03 Feb 2025
By investing in better ways to handle and manage content delivery by IP, organisations can offer greater compatibility instead of locking customers into a proprietary and potentially limiting approach. In the interim, those with IP expertise have a responsibility to ensure interoperability is at the core of decisions that broadcasters are making.
10 Oct 2022
The Virtual Video Control Room enables the management of large-scale live streaming productions entirely in the cloud, connecting remote teams together as if they are sharing a single set
30 Mar 2022