How to Ensure 24/7 Streaming Asset Accessibility in the Cloud

One concern with migrating streaming infrastructure and content storage from on-prem to the cloud is ensuring that your content assets are just as accessible via remote data centers as when you had them all stored in physical proximity on-site.

Steve Vonder Haar, Senior Analyst, Intelligent Video & Enterprise, IntelliVid Research, discusses these attendant challenges and how to overcome them with Rick Capstraw, Chief Growth Officer, Signiant, Richard Andes, Director of Customer Success, Telestream, and Corey Smith, Sr. Director, Advanced Production Technology, CBS Sports Digital, Paramount, in this clip from Streaming Media Connect 2023.

Vonder Haar asks Capstraw, “How do you make sure you have access to your cloud all the time? How do you make sure that the data center is working the way that you want it to work? I know that at Signiant, you deal with a lot of file transfers.”

Capstraw says that one of the biggest challenges is ensuring that file copies and backups are easy to find. In the cloud space, he says, there “Can be a significant cost element to storing stuff that maybe you needed for a quick redundancy but no longer need…the cleanup and archiving of what actually is needed at a reasonable timeline is where the complexity comes in. Amazon's done a pretty good job, as well as the other cloud providers, in providing a lot of different tiering to try to hit exactly the use case that you're looking for. From Signiant's perspective, that's why we try to provide a completely storage-agnostic fabric. So if it's on-prem, it's in the cloud, it's in Glacier, it can provide a way to access that. And then it's really all around automation and workflow of how quickly might you need it back if you need it back, and then that will kind of dictate where it goes.”

Vonder Haar notes that Richard Andes is the Director of Customer Success at Telestream, and he asks him, “Do you get more panic calls on average from cloud customers or on-premise customers?”

“I'd say probably more panic calls from on-prem,” Andes says. “I think, initially, it was more from cloud, but that actually has changed a fair bit. There's sort of two parts to the answer here. If you want some extra redundancy, if you want to make sure nothing's going to interfere with your particular processes, there are options in the cloud. You can do reserved instances. You can spin up two instances, so you have a hot backup available. There are things like that that.” He says that these decisions are heavily based around budgeting. “Is this a high-caliber enough event that I'm willing to spend enough extra money to keep a second encoder instance running or something like that?” he says. “So for live production, you just factor that redundancy directly into your budget.”

Andes emphasizes that some key factors should be accounted for. “How do I maintain continuity?” he says. “In a large fan-out transcoding environment, you're trying to transcode 300 files at the same time. If two of those fail and die, what happens? Okay, the orchestration layer should auto-retry those. It's the same thing if you try to access a piece of media and in one location, it's bad, but there's a redundant copy in another bucket that's in Glacier. Cool, let's go ahead and execute or restore command from Glacier, get that file, use it for the transcode. So that would have to be a relational link within your media management to know that there's that backup copy in Glacier. So it's just a lot of that, making sure that the platform level has the awareness to control and restart, look for secondary backup copies and those kinds of things.”

Regarding concerns about quick restoration, Andes says, “If you are using a particular infrastructure as a service, [leverage] whatever cloud environment you're in, [leverage] what they support in terms of scripting or redundancy controls. So whether or not it's firing off a terraform to spin up four nodes for you or something like that. It's being aware of what's available that can do those kinds of things for you quickly and efficiently.”

He also notes the flexibility of cloud in terms of budgeting based on specific needs. “In cloud, it is a little bit easier because…if you want to have a second version of encode or spun out, cool, just spin up two instances instead of one. If you want to make sure that nothing interferes with your transcode operation, make sure you do an on-demand instance instead of a spot. There are choices you can make if it's worth the money. So I think cloud does open up a lot of really interesting things with that, but it does interestingly loop back to that budgeting discussion very quickly.”

Corey Smith says that overall, there are key issues of both business and operational continuity, which are becoming increasingly complex in cloud computing. “The operators need to know their day-to-day workload or week-to-week workload, what nodes they're going to be actually on, what compute they're going to be on, and keep it all kind of in sync,” he says. “Whether it's updated slates or bugs that need to be [fixed] for that particular broadcast, making sure that the fleet management, if you will, in a hardware-based environment is very easy. I only have so many finite spaces in my rack for hardware. In cloud, I've got five different regions that need the same kind of update and care and feeding.”

Watch full sessions from Streaming Media Connect November 2023. We'll be back in person for Streaming Media NYC on May 20-22, 2024. More details here.