What's REaP?

The term was buzzing around. So I fired off a few questions about REaP to two gents who know more about the subject than most anyone: Mohamad Raad and Roberto Ramos Chavéz, two colleagues at Unified Streaming who are quite active in the research and standardization field of video streaming.

Here are their answers to my questions. Read 'em and . . . REaP. (Sorry.)

Guys, what’s REaP? Just googling it, I know that REaP aims to enable the synchronization of distribution encoder output, such that media segments from different encoders can be used interchangeably. I also know, thanks to Google, that REaP also enables the same kind of synchronization for packagers. And of course, without googling it, I know REaP is referred to as ISO/IEC 23009 – 9. Everyone knows that. But what’s REaP mean to you? Why is REaP important? 

REaP is mainly about building redundancy into a streaming workflow to ensure continuity of presentation in case a source fails. It’s also about storing presentations for viewing later, using a similar approach to VOD.

In reality, customers at Unified Streaming have already been using REaP before it was standardized. We called it Origin and encoder redundancy. Specifically, since Interface 1 of DASH-IF Live Media Ingest Protocol was released, it has enabled running two or more live encoders with Unified Origin for redundancy.

Unified Streaming has put a lot of time and effort into the standardization of REaP to ensure customers can follow a specific set of rules, and reduce the interoperability issues with third-party live encoder vendors.

Okay. So settle this once and for all for me. Is REaP a "guide" standard?

You better believe REaP is a “guide” standard. Basically, it guides the implementer through developing a robust streaming workflow by building redundancy into it. That’s an important contribution to the industry.

How do streaming workflows become more robust by applying REaP?‍

REaP builds redundancy into a streaming workflow, allowing for the replacement of missing segments from one source with segments from another source. To do that, REaP sets constraints on both encoders and packagers to ensure that it is possible to replace segments from different sources.

How’d REaP start?‍

Good question. There was some demand in the industry for what was called “encoder synchronization” a few years before the REaP activity was formally started. This was brought to the attention of MPEG by Comcast in June 2020.

At the time, was the concern basically with the potential failure of the links between the source of mezzanine content, the encoder, the packager, the origin, and the client?

Yes, exactly. Didn't expect that you'd know that. So, as a response to these concerns, MPEG established an ad hoc group to look into this issue.

Unified Streaming demonstrated a potential solution to this problem at the October 2020 MPEG meeting, which helped establish the activity within MPEG as one that may indeed lead to a standard.

Who started REaP?

REaP came out of the encoder synchronization activity. The ad hoc group established at the June 2020 meeting eventually led to the organization of an encoder synchronization workshop in July 2021 for the collection of use cases and requirements from industry participants. This was a rather well-attended workshop, the notes from which were used as the basis for the use cases and requirements document.

At the October 2021 MPEG meeting there was a draft call for proposals based on these requirements. The call for proposals was issued in January 2022. Unified Streaming contributed as part of a joint response from DASH-IF members to this call, and a working draft was issued at the April 2022 meeting.

But what happened at MPEG 139, the online one held July 2022?

Well, what happened was that part 9 of DASH was formally requested from ISO/IEC. 

Up until July 2022, this activity was all conducted under the name of “encoder synchronization.” At the October 2022 meeting in Mainz, Germany—

I've never been to Mainz.

Okay. How's that . . . okay.

Is it nice?

Yes, it is. Mainz is nice.

But is it, like, super nice?

Depends.

On what?

Doesn't matter. So at the meeting in Mainz, the main participants in this activity decided that the name wasn’t really suitable for what was being developed. So, on one sunny afternoon during the week (it was either Tuesday or Wednesday) that group came up with the name “Redundant Encoding and Packaging.” And so REaP started.

Where’s REaP now? What phase? 

REaP is under FDIS ballot. The ballot closes on April 18th, 2025. If approved, then the international standard should be available by the end of June 2025.

Who wants REaP to happen, besides you two?

After the first workshop in July 2021, and the successful balloting of the committee draft (CD) of REaP, a second workshop was held in July 2023, where the draft international standard was shared with participants and feedback was collected. That workshop was also generally well attended, showing that there was still interest in the standard three years after the activity was started.

Where’s REaP gonna go?

Encoder and packager vendors will see a benefit in using REaP, so it will probably go into products.

Where’s REaP not gonna go?

I doubt that REaP will be expanded, but it may be. We’ll see.

What’s the ultimate goal of REaP?

To enhance the quality of experience for streaming viewers.

When will REaP end? Will REaP ever end? Should it?

I doubt it will “end.” This is not an encoding technology that will get replaced by better encoders, for example. This is a way of making your segments redundant by accurately labeling those segments.

If REaP were a meal, what would it be? Something hearty like a steak, or something more like a side that sorta rounds things off nicely? 

It’s really something that allows you to have your meal from start to finish, with no interruption.

So, fast food?

REaP is good for you, unlike fast food. The food metaphor, that's difficult, it doesn't really apply here. But REaP's basically something that helps you finish your meal in peace.

So, like a digestif. Sorta like a limoncello, an amaretto, or a brandy, even.

Hmmm.

But this is great. Thank you, guys. Is there a link I could go to, to understand more visually what REaP's all about?

Sure. Check out https://github.com/unifiedstreaming/live-demo-cmaf. That's where you can witness a live ingest demo, sorta the historical place where things started with REaP. Get to the root of it all.

I would totally check that out, but my browser's got too many windows open, and it's super slow right now.

Not our problem.

It's not?

Huh?

Okay, thanks!

Any time!