How A+E Networks built a cloud-based media supply chain for scale & agility

All right. Hey, good morning folks. Um thank you for coming American Pickers 1st 48 Storage World Wars. This is A&E Networks and today we're going to be talking about A&E and what they've been doing with AWS in the cloud for the last five years.

My name is Brenda Noone. I'm a District Sales Leader out of our New York office focusing on our media and entertainment vertical. I'm here with Tanya McKee, Senior Account Executive for media and entertainment. And today we're excited to welcome our friends of A&E Networks to the stage.

Uh we've been working closely with A and a now for the past five years, they've been transforming their business. Um they, they've been looking at starting early in the cloud from lifting and shifting their environment to modernizing their environment and now working on across different lines of business when it comes to their media supply chain. So we're here all about their journey today.

So many of you have likely heard of Lifetime FY I Vice or the History Channel. But what you don't know is that A&E Network is much more than that they produce content for large distributors under for their unscripted content. Six West, uh Category Six and A&E Studios for their scripted content. So how many of you have watched Lincoln Lawyer on Netflix or Big Sky on ABC? Well, that's A&E Studios and we are going to have two people from A&E presenting with us today.

First Don Jarvis, he's a Senior Vice President of Global Broadcast Operations and Engineering. Um, interesting facts about Don. He was on stage with Madonna once before. He's played craps with Bruce Willis and he spent time in a bathroom with uh Billy Idol. I think it was as well at the MTV Video Awards. So he's had, he's been in media and entertainment for the last four decades and has a lot to share with us today as well.

And we'll also be introducing David Clee to the stage who is never shy of a challenge and actually a hike. the Matterhorn. Very, very cool. Um and I'm really excited to, to share the story today with A&E we've been on this journey with them on reinventing their supply chain for the last couple of years.

But before we get into it, we couldn't have a media and content session without you guessed it. Media and content sizzle time play the track.

[video plays]

As you saw from our sizzle reel, we produce and distribute a fair amount of content globally. That's right. And uh I'm gonna go rest my voice for a few minutes, but when I come back, we will talk a little bit more about what we've done to change how we do that over the last few years, building our cloud based supply chain in AWS with a lot of great partners with that. So I'm gonna take a few minutes to go through our journey and it is a journey. There is no destination. That's a continuous journey uh from where we started to where we are today.

Uh then David, as he said, he'll come back and he'll discuss the architecture and use cases as well as the companies that we partner with in AWS. Well, then give you a peek at what we're uh, have um slated for the future and uh we'll go ahead and uh end with uh some q and a if we have some time. Does that sound good to everybody? All right. All right.

So, uh I know most of you would like to go straight to those architectural slides, but there is no skip recap function on this presentation. so we're gonna let david go over there and rest uh while I catch you up uh to where we are today.

So we started back in 2016. Um and we were looking to see what the next technology shift for media entertainment was going to be. We had already been through the analog to digital conversion tape based to file-based sd to hd. So as we do in this industry, uh we reach out to peers, we discuss common needs and solutions, successes and failures as well as attend industry events such as this one where we talk about our shared experiences.

Uh throughout the year, we heard presentations by chris blandy on nat geo's migration and chris now is uh with the m and e vertical within a us. So, thank you very much for that, chris uh as well as mike cotter of uh turner media, josh urby of discovery networks. Um these success is validated that software as a service and platform as a service running in aws were viable solutions for media centric processing.

However, for a&e to take advantage of these a business case was needed to justify that move to the cloud. In 2017, we continued to follow developments on media supply chains. Our internal multiyear cloud city initiative was underway at a&e with our technology team migrating our enterprise systems applications and infrastructure to aws. So we already had a solid cloud partner. But the business case for cloud processing of media was still not justifiable.

In 2018, 2 issues converged to provide that justification. Uh the existing technology was on premise, purpose built for supporting our traditional broadcast linear channel business, meaning it dealt with processing our premiering content and delivering it to broadcast as well as to our mvp ds for provided services. So we're talking about a few 1000 assets a year distributed to a fixed number of outlets.

Then we started to see some small library deals come through. Uh when we started to see these, that meant that we were going to double the amount of assets needed to be processed. So now we had a fixed infrastructure that was oversubscribed and was also slated for replacement in the following year. The time was right for a a to do a technology shift.

David Clee was brought on board to provide the strategic leadership on solutions. As he started down our happy path here, our business case received financial approval after which we ran an rfp which resulted in a two pronged solution. We would outsource to a service provider for short term immediate needs and we would begin developing our cloud based media supply chain for the longer term needs of the company s tb i rally was selected as the workflow orchestration tool to run an aws.

And for the remainder of 2018, we built a dedicated team within engineering and began developing on the platform we called media elevate. This is our internal branding that differentiate us uh for media process from all the other cloud initiatives that were going on within the company, we iterated on the name came up with a few verticals that the workflows would fit into

We laid out an engineering focused road map with initial workflows developed to receive finished assets into the platform for processing for broadcast, with additional workflows added in subsequent years.

We built a confusing status matrix of content types versus process with too many check boxes and swim lanes which made cross-company collaborative communication ineffective. So we adopted our terminology called "in" and "out" so that we were consistent when working with our finance, technology and operations teams. Cross departmental coordination improved.

And just recently, we started to layer back in some of that granularity when we start having road map discussions.

In 2019, we addressed the over subscription issue of the on prem plant. And by April, we came out of development and launched Media Elevate into production. We were moving down our happy path.

Our initial success was based on supporting our linear channel business and the associated volumes - a relatively fixed number of new assets each year developed to a fixed number of endpoints on a regular cadence.

Almost immediately after launch, we needed to pivot - multi platform became a huge component of our business. In a very short timeframe, we had streaming channels, FAST channels, AVOD, TVOD, D2C applications, international buybacks, internally produced original content, short form content, podcasts - all requiring media processing of some type.

The platform was expanded, adding new workflows to meet these dynamic new business opportunities to support the workflows. Our existing 30-year library had to be available for distribution. So many of you, if not all of you, know the term "lift and shift" and how you should not do that when moving your workflows to the cloud. Otherwise you don't get to take advantage of all the opportunities that the cloud provides to you.

For those of you who have not started your cloud migration - here's my one takeaway for you: Do not lift and shift your content library into the cloud. Our company spent over two years doing the migration and it was a huge cross-departmental project from our technology department setting up new data streams to legal examining rights and clearances to our operations team pulling our content out of hard drives, LTO libraries, video tapes, cleaning them up, adding metadata to create a library of golden assets that are now stored in AWS and can be used for future distribution.

So for the remainder of 2019, we scaled up to handle the increase in new content being received and processed in Media Elevate by bursting when necessary to support the library digitization project and having provided secure remote access for external service providers acting as burst operations.

Things were progressing nicely as we moved into 2020. We took advantage of the agility and elasticity of the cloud as we moved forward further down our happy path with new workflows and increased scale.

And then we experienced another unexpected pivot in March 2020 - we were asked to leave our purpose built technical spaces and continue processing at the same pace from the comforts of our homes. We were pushed off our happy path to a bit of a rocky road. But our cloud based media supply chain minimized the chaos, allowed us to handle this disruption in a way our facilities based systems never could have.

Agility and elasticity were the two principal advantages we leveraged throughout this migration - from financial approval to design and implementation. However, accessibility proved to be the most valuable in our transition to remote working and continues to be today in our hybrid work model.

It was a scramble, as I'm sure it was for many of you, but we were able to modify our workflows, send laptops home, change our support model from supporting one centralized technical facility to supporting a diverse collection of residential locations. We met those new needs in short order and I'm happy to report that we did not miss a single premiere across any of our distribution outlets.

For 2021 we continue to iterate on our "in" workflows - library digitization is processing. So our S3 buckets are filling up nicely with content. We have enough volume in Media Elevate that we begin building the "out" portion to fulfill large volume bulk deliveries to some of our endpoints.

In 2022 iterations continue on workflows already in production and development of "during" begins with proof of concepts on cloud based editing, focused on the formatting, fixing and versioning needs for distribution of our finished shows.

The operational requirements are locked, technology has been selected and migration of the existing services will be completed by mid 2023.

Now, I understand how we do these wonderful things at a 30,000 ft level, but David Clee is gonna come back on stage and he's gonna explain them to you at more of an atomic level. So I'll now turn the presentation over to David.

Thank you.

Hopefully won't go to the actual atomic level, but happy to get into a few more details on what we thought about the build, how we approached it, and then each of those phases - in, during and out - what we built for each of those phases, why we did it that way and what's currently coming online.

Those three phases, those are really derived from traditional supply chain manufacturing ideas - the idea that roughly you have an input at one end of a supply chain or a process, raw materials come in, you put them together and then you have a finished good at the other side - a refrigerator or a car or a microwave that you're ready to sell to somebody. That general principle actually applies pretty well to some of the work we do in media processing too.

Credit to STV I who were some of the first people that I heard kind of talk about that in that context, at least for this supply chain processing for media. So what we do in media is not that far off in some ways - we take in raw materials, we might have a finished master of a show come in, maybe some extra audio tracks, some captions, some images, some metadata, we have to put that all together in the right way, make a few different finished goods out of it. And then in theory, those finished goods are actually seen by people, consumed by people, have value - something that people want to watch.

So this really has become the foundation of how we've approached our media supply chain build at A&E - leveraging some of those traditional supply chain principles for manufacturing.

As we've done that, we've taken what has been a relatively straightforward process and remapped some of our supply chain topics into these three phases - this "in", "during" and "out" that Don referenced earlier.

So "in" being the phase of the supply chain that's all about getting content in and making sure it's good quality to be used downstream. "During" being all about the mixing it up and making changes to it and getting it ready for downstream use cases. And then "out" is all about getting the content out and getting it distributed to all the downstream partners that we need to send to.

When I started at A&E, like Don mentioned, I was new, starting a new group from scratch and had a relatively unique opportunity to kind of make my way around the organization and meet people and find out a lot about how this was being done today and what people liked and what people didn't like and what the legacy systems were doing well and what we needed to improve on or see as opportunities.

And there were three really big themes that came out of that:

1. Agility - As a business, we wanted to be much more agile. We wanted to be ready as a company to jump on whatever the next big thing is, if it's a new type of content, a new type of distribution outlet, some sort of content topic that gets hot for a while. If it makes sense for A&E we want to be there and we want to be there at scale.

2. Elasticity - The other thing that was a big theme, especially with our operational teams. Everybody was really tired of waiting in a queue, waiting for their stuff to process, waiting to see if their episode of television was going to go through to where it needed to go today. Everybody talked about the idea that wouldn't it be great if we could ramp up and ramp down and just get rid of the queues and get everything through as quickly as we could.

3. Efficiency - The other big thing, as Don really referenced, a lot of these media processes were designed to basically do one thing and they were designed to get content to master control playback. They were designed to get content playing back on cable or on broadcast. And now it's a couple of decades later and a lot of these processes are being asked to send content hundreds of different places. And the complexity is really through the roof on the different types of content, the different things that need to be done. So everybody has also seen an increasing of the staff to meet that complexity and to meet that need. So everybody talks about being more efficient, doing more with less, trying to make sure that we can do some things to be more efficient in the way we work so that we can take on all of those new things that are coming through the door.

So based on those three pain points - for agility, we talked about a design principle of putting together tools that were loosely coupled and targeted to do specific things. We're not trying to build any monoliths. We're not trying to solve all of our problems with one big system.

For efficiency, of course, where else do you get that other than the cloud? Where else could you look at ramping up not to just do like, you know, 10 times the normal day's work, but maybe 100 times a normal day's work or 1000 times a normal day's work without having to build that all out?

And for efficiency, we talk a lot about lean manufacturing and those same general principles of manufacturing supply chain and some things about that I think are very good fit for media supply chain:

1. It forces you to think about the process at a very high level and think about the steps that you're doing. Do all those steps make sense, do they all actually add value to the content, to the process, to whatever you need to accomplish?

2. It encourages you to think about how the steps flow together, that there's a lot of inefficiency in the handoff between steps of media and materials flowing from one group to another. And focusing on that and focusing on making those handoffs more efficient actually can give you your biggest bang for the buck maybe even more so than making each individual process as efficient as it could be.

So with that, we got into building and for each of these phases, I'll go through a goal at the top - kind of like a prime directive or raison d'être, the reason why this phase of the supply chain exists - some things that we do to help accomplish that goal and then some techniques and tools we've built along the way and found useful.

And then we'll go into an architecture slide for each of these three phases - in, during and out.

So for "in", the goal is obviously to get the content in but not just any content, content that we actually think is usable and is what it says it is and will let us do what we need to do with it.

There are a lot of steps here upfront in data entry where you need to make sure people have told you what you're getting and why, and you know, a little bit about it, you know, a little bit about what its intended use cases are and you're able to line up a specific delivery where somebody's uploading content to you to that information and that metadata that you have that's supposed to tell you how to use it.

And then we've been building in a lot of automation in these cloud based processes for ingest to help validate that the content is what it says it is - check its header, scan for viruses, scan for corrupt frames, things like that.

And then almost everything new coming in the door, every, almost every new episode of television or movie ends in a human review. And so for us, it's been very important to make that human review step as efficient as possible. And that's sort of where I'll jump down to this efficiency section here.

So much of the time the human review historically had been involved of marking things from scratch and kind of assembling information from all over the place in order to get exactly what they need to get done. We're trying to do a lot of automation ahead of time and make the computers do the things that they're good at and then let the humans do the things that they're good at and confirm if the computers did things that made sense or not.

Um so think of it like a computer is very good at those objective measures of is this the video format we need or not? Is this picture too bright or too dark? Are these audio tracks too loud or too quiet? So let the computers make those decisions and then present that to the user and then the user can review that and see if it's right or wrong. But then more importantly, add in their part, which is the more nuanced decisions, they're much more subjective, much harder to get a computer to do and really make more sense for a human to be involved with. Like, is that violence on the screen? Ok, for a tv, 14 rating or is that picture in the background of a show? And an important part of the story needs to be translated and put on screen for people. So it's really get the automation to run and present it to the human reviewers and the human technicians in an efficient way, right on a timeline. So they can see it in place and make decisions quickly about what needs to be done when it comes to elasticity.

Uh we've absolutely found that uh pinning up infrastructure is great and we'll talk a little bit more about that in a bit. Um and the thing that i think maybe i didn't anticipate was how useful s3 would be for us. We're using an awful lot of s3 and it has been awfully nice to not have to think about the content going into the lto library and expanding our lto library and thinking about restore times and externalizing tapes and all the things that go with that physical infrastructure maintenance, particularly for larger archives. So we're just pushing things in s3 and it's scaling for us. And it's been really nice to have that way when it comes to agility.

Uh we have had the modern software design principles of asynchronous messaging between targeted tools be very valuable. But i think the bigger thing here is the idea that uh we realized our users could work from anywhere that they had access to the cloud, which was pretty much anywhere. And that wasn't necessarily what we went in thinking was the priority. But as don mentioned, some things happened that made that a very important priority for this build.

So let's talk a little bit about how this all comes together. Um I'm gonna start with this editor over here on the left of the screen. This is an editor at a production company who's finishing out their episode of television or the movie that a&e has asked them to make and they're uploading it to us through debut, which is a homegrown web portal made by our very capable technology group. Uh they have have enabled a web portal where people fill out metadata have the information they need for that submission and then upload the file and the file lands straight on through with espera in our in just three bucket. And from that point on this is where that asynchronous messaging starts to kick in and be extremely valuable. We're notified within our systems that the content has arrived. We're able to notify std. I rally a workflow engine, the content has arrived, it kicks off some workflows. It's in turn sending us back messages about the workflows, the progress every step at a time, we're back on our end making updates to other systems telling other systems internally, even corporate systems that hey, this piece of content has arrived and hey, a screener is available for it. Now, if you'd like to take a look each of these pieces, this a asynchronous messaging has been really valuable to expand and to scale as well.

Um but over on the, the far right, what you see is the users primarily getting their work done from a web browser. And that's the thing that uh i think we all found extremely valuable as we got going, whether you're a user at home or you're in an office or you're a third party because we got really busy and we needed some more help. So we brought in an external contracting company to help with some review and approvals and previewing of content. You can get an awful lot of your work done with a web browser from anywhere. And we do have some adobe premier based uh review processes. I'll talk a little bit more about in the editing section, but most people stay in a web browser most of the time and get their work done, which has made it really easy for us to scale this process and to make it accessible as we've been growing. And when they're interacting with the process, they're interacting with uh our oms our operations management system, they'll talk about in a minute, gives you more information about what work needs to be done and why. But they're also interacting with stv i rally, which is our central workflow and orchestration engine.

I want to spend just a minute talking about that. Um stv i rally uh is a third party sas tool that we use. This is uh i asked them for an architecture diagram. This is what they gave me just to have a little look at what's going on under the hood there. But this third party sas tool does a few key things for us to enable the things that we wanted. Our design principles in the cloud. Number one, it's a workflow engine. It's where my team goes in and actually programs in the business logic and the decisions of what's going to happen when. So uh a new file of this type arrives, we need to make a proxy of it, check this format, notify these people and then send it down the line to a review process, those decisions, those business logic steps are coded into rally and then rally is able to help be the state machine that keeps track of them and helps be able to monitor across the way infrastructure automation though, this is really where um we're able to leverage that elasticity of the cloud.

Uh see if this, this uh this kind of works. Um right down here, these things are called provider instances. This is one section of the um diagram for the stv i rally tool. That's where the heavy lifting gets done. That's where things like transcoding, one video file to another uh scanning a video file frame by frame, doing a check, um doing a virus scan those provider instances, spin up when they're needed to do work and then spin back down when they're not needed anymore. Uh and that's been great for us. So essentially, you know, if we have a new file coming in the door, it needs a trans code as a part of its ingest workflow. And maybe that transco uses a tool like a tre v. The file shows up the workflow kicks off a rally spins up for us. One instance of vantage processes the file file is done and the vantage instance spins down 50 files show up in the s3 bucket, more or less 50 vantage instances, spin up process all 50 files, they all move on in the workflow, then it all spins down. It's really given us that elasticity and that ability to avoid having cues and people have really enjoyed that.

Now what's running on those instances is a mixture of some software that we've gone out and licensed ourselves, but also some software that we're acquiring on demand through stv is application marketplace. So they offer, uh i don't even know how many now maybe 50 different applications. I'm, i'm making up a number that you can leverage through the application marketplace and just pay by the minute for. So we have some tools where we're not sure if that's the right tool for us or it only works for one particular workflow. We just want to try it out. We can basically work it into a rally workflow. Have that spin up. We pay by the minute to utilize that software license and the infrastructure and then it spins back down and we're not paying for it anymore. That's allowed us to be really flexible, trying out new tools. You know, most of our workflows mix and match transcoder, they're good for the job. Now, we have maybe six or seven different ones we use regularly. We just can put in the right tool for the right job. It makes it really flexible. We can try something out for a while and can kind of rent it. And then if we really like it, we go buy it, we go make a good deal with that company and try to work out a long term agreement.

Uh and then the last thing in raleigh uh of note to me is the web interface lets you track state, lets you get a good sense of what's going on and uh and it's been incredibly valuable for us as people are monitoring these work flows from home. So uh that elasticity and agility i talked about has been really helpful for ingest.

We started as don mentioned, kind of planning for our normal day to day new content ingest. And that was about 500 masters a month. These are high quality large files, but we're not talking about uh a giant amount of content coming in the door every month. That's a good amount. But then pretty much right after we got it online, our major initiative of library digitization kicked off and suddenly we had to now handle getting another 1500 files in the in the door every month. So about three times more than what we planned. Uh and then of course covid hit and we realized, wow, we already have this web-based review and ingest tool that lets people work from anywhere. How about we get those last few broadcast workflows that we need out into the cloud and try to help get people out of the building, stay safe and facilitate working from anywhere. So that added in another uh 3000 broadcast segments a month. So at the end of the day, the system that we designed uh is doing about 400% more work than what we planned it for.

Um and from an infrastructure perspective, we really had to do almost nothing to make that happen. Like we didn't have to go buy more servers or buy more storage or adjust our networking, the the processes themselves just scale the infrastructure in this scenario, almost just sort of fades away into the background. And it lets our team focus on the more important things of individual unique business logic, making things for the business users to let them work the way that they want to and help give us an advantage, doing the things that are unique to us.

You can kind of see on the chart here, this is um monthly in volume and you can kind of see we were planning for a workload, you know, after an initial and just down there at the beginning, then some other things kicked in and now our day to day work loads uh going out like i said, about 500% of plan

Um and it really was a great experience for us to just see it scale and just be able to move on and keep building new workflows, keep working on to build whatever the next important piece of business logic was.

Ok. So let's go on to editing. So editing, as Don mentioned is the newest of our um phases that's moving wholly into the cloud. This is supply chain editing. The goal is to make changes to the content for linear and nonlinear distribution needs multi platform needs uh every new master that comes in every new finished episode of television, usually needs 3 to 5 different versions made during its life cycle in order to be useful everywhere it needs to go, we might need a slightly longer version for a premiere, maybe a slightly shorter version for a rerun that has um uh one more commercial break in it. Maybe you need a version without a text on the screen as a base for international distribution. Um maybe need a version with all the commercial breaks and all the bumpers and teases and promotional content removed.

Um each of those things gets done today by a video editor. Um so here again, this theme of analyzing the content up front as much as we can putting in a lot of automation. So that when the video editor gets involved, the video editor has all the information they need to do that work in front of them, knows exactly what changes they need to make. Those changes are up for them on a timeline and you can get their work done quickly and then do some annotation, tell us about what they did so that we have good metadata to automate downstream work flows out of it.

So that flow that we started up with for our review process, we've actually really tried to build out and leverage. We found it very successful for these editing processes, run as much automation as possible, get it ready for people to get engaged, maximize their time and then automate as much as possible afterwards.

So let's look a little bit about how this comes together. Um this time our users start on the far, right. Um again, they can get most of their work done through uh vm. They log into the web browser, they can see what work they need to do and some details about it in our operations management system, they could see some status information and rally. And then when they're ready to edit, they're using terri dhi uh the uh pc over ip client that's gotten to be pretty popular lately, taichi to log in to an actual editing instance, an actual ec2 instance and that ec2 instance is set up with all the right software for them. Adobe premier installed all the right settings, all the right presets to enable this type of supply chain format editing format transformation work.

And this rally access panel installed again from stv i, something that talks back and forth to the supply chain allows us to take data that was entered upstream, put it right on the timeline as markers, let the editors do their work at those moments in time, add some more markers and then it goes back up and actually renders back in the cloud. And that edit environment is all set up for us with this new platform, this uh relatively new partner of ours arch platform technologies, they basically build post environments in the cloud, an awful lot like a post environment like might look like on the ground. You know, you have computers, they all attach some shared storage, they all have their settings tweaked and optimized to be using adobe premiere and to be interacting with the services that you need and you have the presets that can follow people around. It's really, it ends up just being a pool of computers as a user. You don't even care which one is free. You just pick one that's free. Your information follows you around. You log in, you have your settings and you have all the settings that the system has designed that are going to be optimal for you to do whatever particular job you're doing.

Ok. Um so here again, the theme of automation is big in play. We're trying to do a lot of automation to prepare work for the users to get engaged. Uh and one thing i want to point out is a new partnership we have with uh prime focus technologies. They worked with our operations teams to actually develop uh this machine learning algorithm that was able to help automate some of our more repetitive work inside of a very common editing workflow for us.

Um and that was part of our library digitization project where we're pushing through those thousands of titles. Well, we've been able to work with them to leverage that as an api. So we can actually call their api point them to a piece of content that we want analyzed. The service will analyze it for us and then send back to us an a i edited timeline, a robot edited timeline that we put right into the premiere panel for the users. So when they open up their job, they see that pre edited timeline right there ready for them to go.

So then the users go in the editor um confirms and denies what the a i has done and then has to do some additional work um in our addition or original proof of concept with this earlier this year. Uh the editors said they saved about 50% of their time from doing it from scratch versus doing it with this a i edit assist doing the work ahead of time for them big savings. And we're really excited to see not only where that a i goes, but where the rest of this editing workload into the cloud goes. As don said, it's all coming online between now and april.

All right. Now, our last phase distribution um the point of distribution, the goal is to deliver content and, and materials to end points to take all the stuff we've been, we've been working on and get it to where people are actually going to watch it and see it.

Um here again, uh we have human and automation involvement, but the human involvement actually comes up front. We have uh a lot of people going through to take care of schedule information, uh when things are promised to who and why, um, wrangling information from many different corporate systems to try to make sure we have a very clear set of deliverables that all happens up front. And then from that point on, we really want the rest of the process to be as automated as possible. Once somebody puts in the schedule, somebody puts in an order to deliver a piece of content, we want to have almost everything else from that point on be automated. You know, whether it needs to be a longer or shorter or insert graphics or remove certain audio tracks over others. Everything from that point on, we really want to automate as much as humanly possible. It's really we found that's the key to speed.

Um even if you have like stages in your workflow where you're used to having a human check a box to say approved, send, ready to send, approve. If you're sending thousands of things in a day, you can only click your check box so fast. And so you have to really think about which of those steps add value and what you can do to make the process as automated as possible.

So let's take a look at what this looks like at a high level. Again, it starts with our users, whether they're at home or in the office, they're doing all this work through a web browser. They're interacting primarily uh with our web based tools stv i rally and the o ms.

Uh and this is a good time to talk a little bit more about this operations management system. This o ms called polaris. Uh we've been building this with our partners at tmt. This type of system you've seen on the previous slides is involved in all stages of our workflows. It's where people keep track of how things are moving across the supply chain, understand the status and the state. But it really shines in distribution heavy workflows like this. This is where people can go in and create the concept of an order. So if i have 300 shows to ship to a particular destination by a particular date, you can go into o ms make that an order in the system and then you can you know, ask that order questions like do you have all the materials you need? Uh are all the masters here that are in this order? Are they all already in the right format that they're supposed to be in? Do they need rendering? Do they need additional information? And then most importantly, have they actually gotten where they were supposed to go on time? Is the order complete, is the order fulfilled.

So by treating things as an order and building up that kind of uh workflow, we're really hoping to streamline interaction with the supply chain and give people a lot, a lot more valuable information about how things are progressing.

Um also in this stage, we found that the integration pieces are super important and i haven't talked much about these yet, but you can see down on the bottom, right. Uh we have as pera servers that's a popular uh uh you know, w acceleration tool. It does point to point communication. We use many other tools as well, things like sign and, or even ftp. In some cases, the idea that we can set up a connection to another system or building no problem in the cloud, we can run those servers and set up those connections to other outside destinations. But what we found it to be really powerful is when we can set up direct connections to other people's aws s accounts.

And as this has been getting more and more common lately with more and more media companies moving a w us, there's kind of a a network effect kick again where we can talk with somebody about setting up a large content migration and that large content migration can be enabled by a direct s3 to s3 transfer. We sort of have built up a cookbook of three or four different ways. We like to set this up on a regular basis. But when we can get that transfer happening entirely s3 to s3, the speed is just through the roof. It, it totally takes uh away a concern about speed and transfer times even on very, very large orders. And very large assets.

Um one story about that, i'll tell you as i show some of our delivery statistics. Um so early on as we were building up distribution in 2020 don mentioned, um we had a fairly strategic deal come up with another streaming service. Um uh a&e has, is not uh has strategically made the choice to not have a plus, at least right now where uh we are reclaiming all our content, we instead uh have the opportunity to distribute and make partnerships with a lot of other organizations that do have a plus and help them round out the value of their library and provide additional value to their subscribers.

So one of these deals came up fairly early on and it was a fairly large volume deal with our operations team and the legacy systems. They said it would have taken 30 to 60 days to complete by a team of five.

And basically those people would have been sitting there, you know, kind of watching paint dry for, you know, a couple of months like click upload, fill out the form, wait for the thing to go click upload, fill, you know, for a while.

So we were able to work with the engineering team at this other organization. It was fantastic uh and able to set this up as a cloud to cloud transfer where they were also a Rally user. So it was really from supply chain to supply chain and we got that 60 days worth of transfer done in about 16 hours including all the renders including the transfers between S3 to S3.

It was a real eye opener for us to see what was possible when everything was set up for automation and ready to go. Um and uh if you look at the chart, uh if you can, if I can kind of use this, this one of these smaller blips down here was one of those uh earlier that process, that story I just told and people have asked again. Oh, that's interesting. Could you, could you do that for this deal? We have another thing coming up. Could you do that again?

So you could see the blips have kind of kept going and kept getting bigger by and bigger. This is delivery volume per week. Um and uh even though it's per week, I, I happen to know this tallest number here, this tallest point in the chart, we did that delivery in almost one day, uh pretty much one day and it took uh processing about a petabyte worth of master content to get that delivery done in that one day.

And I just don't know how you do that without the cloud. I just, I don't know how you process that kind of content, that much content that quickly without these tools available to you to use. I mean, imagine if you're trying to restore some of that out of an LTO library or you're trying to get it off of externalized tapes on shelves. Like, I don't know if that deal even gets done at that point.

And that's what I think is really special about this is we're, we're trying to enable new things for the business, new ways of working that people haven't been able to take advantage of before and hopefully help take advantage of some new opportunities that otherwise maybe we wouldn't have been able to.

So coming back to our design principles where we started, um we felt good about these, we feel like we were able to build cloud native all the way through leveraging these ideas of agility, elasticity and efficiency. But we also realized we left a couple out. Um and as Donna sort of referenced these too authority is the first one, metadata, good metadata governance. Great cataloging the idea that if you want to have information automated as much as possible, you really need to have metadata locked up.

And the idea that uh targeted systems that are loosely coupled together have to agree on what they're calling things. If there's a lot of data transfer going on in these targeted systems, each system has to be aware of what field is, what and what you call this thing and how to reference this other thing. So having good governance around that process is extremely, extremely critical and accessibility was the last one that while building in the cloud, maybe wasn't uh a point of ours to go work in a web browser alone and work from anywhere.

You know, when we started this, we were all in the office and that was pretty much fine. Um it's now very much a point of the process that we want to be able to work from anywhere. We want to be able to do that work from a web browser. So making interfaces that are accessible to people and also making accessibility being that they're really easy to use.

We want people to be able to get the information they need quickly. You know, is my stuff here yet. Will it get where it needs to go on time? These are all very important pieces of the puzzle that we want to make it easy for people to understand.

Last thing I'll say before handing back to Don is that kind of these are the key takeaways for me. At least in this process is Don referenced very little is lift and shift here. It's all rethinking processes, rethinking software design and even rethinking the content and how it's structured and how it's annotated. In order to get this done metadata, metadata really is the fuel for automation cannot have great automation, cannot have this great scale without great metadata.

And coming back to the idea from the beginning of manufacturing this flow over function, I really do think it's true that you get more bang for your buck. You get a better outcome, focusing on the handoffs between the steps and all the things that can get lost in efficiency at those points rather than focusing on each individual step and trying to make that as efficient as possible. When you all work together, when you all are involved in one supply chain and everybody's focused on the flow from end to end, you can get people thinking in new ways and actually get some really great things accomplished.

Uh so with that, I'll pass it back to Don to talk a little bit more about where we're looking, moving forward. He's really good, isn't he?

So thank you, David. Thank you for soldiering through. We know he, he, he lost his voice last night. I was a little nervous that I was gonna have to take part of his presentation which would not nearly as been as informative as having David do it.

So, uh so looking forward, anybody that's been around uh the media entertainment business for the past four or five years knows how dynamic it is right now. Nothing is etched in stone. Uh being in AWS has allowed us to use the agility elasticity and accessibility of the cloud to meet all of A&E s uh future media media processing needs.

So, we're expanding from that happy pathway that we had. We're now on a super highway and in 2023 we'll continue to iterate all the existing workflows to support new technical or business requirements such as workflows for higher resolutions or international languaging.

Uh the OMS that David spoke about will uh continue to be developed to provide ease of use for all the operations teams. We'll complete our migration to the cloud uh for edit of finished assets.

Um as enterprise systems are augmented or replaced, we will partner with our technology team for higher and tighter uh interconnections with ME to elevate to facilitate better metadata flow and opportunities for uh automating more processes.

We'll be uh defining the requirements for integration with enterprise asset management systems for both productions and finished assets. This includes work flows for the acquisition of source material.

Uh expanding our use case with our partners like uh Prime Focus of AI for uh process and additional content types such as alternate language, correct uh creation.

Uh and then David spoke about a couple of the success stories. We have the interconnection of supply chains, right? So these are uh the interconnections within our company are are important, but we think there's a lot of opportunities in the interconnections between us and other media companies.

Um we think there are some big benefits here with regarding efficiencies, we reduce redundancies. Um and as David spoke to, we've done some work in this area, but we think there are more opportunities.

So if you do business with A&E and you're not in AWS, come talk to us about the benefits of being AWS. All right. If you do business with A&E and are in AWS, but don't have a media supply chain. We'd love to hear from you what your future plans are. And if you do business with A&E and have an AWS based media supply chain, we need to talk to you about the benefits of interconnecting.

Done. Yeah. And I think the last thing I'll say, we really do believe in collaboration at A&E. It's, it's a really important part of our company mantra uh just across the organization.

Um and we've had some great collaborators within our organization, some great ones here on screen. This is our core media supply chain engineering team, a group of people who I'm really proud of have done a fantastic amount of work in a pretty short amount of time.

Um and uh this has sort of become our, our makeshift mission statement, our mantra that we keep saying make the complex, simple and make the simple automatic. We're not really sure where it came from. I actually think Eric came up with it, but like we don't know if it can be attributed to somebody. So if you know, let me know, but um this has been a good, become a good mantra for us because it puts the emphasis back on the process thinking through why the process happens the way it does understanding it kind of understanding how we're gonna add value as much as possible.

Um but to a person. This team has been a great set of collaborators, even though many of them are new to the organization, they've been able to learn a lot about how things work in a pretty short amount of time.

Um and we've had some other great collaborations internally to our, our technology team who's a couple of folks are here with us today. Uh Kevin and Jeff uh had the Cloud City project that really paved the way to the cloud for A&E pretty hard to go wrong following that. That's a right up on the AWS blog that I encourage you to check out as well.

Uh when you're already having your organization moving that way to the cloud coming in after that with the media uh opportunities makes it a lot easier.

Uh and our operational teams too are always so busy, they're always working on so many things to try to keep up and they always make time for us to work with us and to try to come up with new ideas and be collaborative.

Um so we've had a really a great run. I think we've, I think Don and I would agree, we've been dealt a really good hand at A&E when it comes to being set up, right, to be able to do some big things and to make some changes.

Uh and I think we're both excited to see where things are going to go and what we'll be able to do over the next few years to help continue to make us the most uh agile media company on earth.