AWS re:Invent 2022 - Sky’s journey to migrate live sports and news to AWS (MNE203)

Aug 16, 2023

AWS re:Invent 2022 - Sky’s journey to migrate live sports and news to AWS (MNE203)

In this session, Sky will share their journey to recovery from a severe fire accident just in time to face a new challenge: the pandemic. These two challenges inspired innovation to boldly redesign Sky’s broadcast environment, from production to playout to distribution, on AWS. Using the flexibility and reliability of the cloud, Sky remodeled their internal ideologies with startup methodologies, driven by a growing diverse team, to deliver their long heritage of compelling events within an agile framework.

Learn more about AWS re:Invent at https://go.aws/3ikK4dD.

Subscribe:
More AWS videos http://bit.ly/2O3zS75
More AWS events videos http://bit.ly/316g9t4

ABOUT AWS
Amazon Web Services (AWS) hosts events, both online and in-person, bringing the cloud computing community together to connect, collaborate, and learn from AWS experts.

AWS is the world’s most comprehensive and broadly adopted cloud platform, offering over 200 fully featured services from data centers globally. Millions of customers—including the fastest-growing startups, largest enterprises, and leading government agencies—are using AWS to lower costs, become more agile, and innovate faster.

#reInvent2022 #AWSreInvent2022 #AWSEvents

Content

0.12 -> - Good afternoon everyone,

1.2 -> and welcome to MNE 203 session about Sky migration.

5.46 -> This session is a narrative about how to face

8.16 -> the challenge of the transformation.

10.59 -> We know that the broadcasting industry

12.33 -> is a mission critical environment and we are so focused

15.9 -> to warranty the business continuity

17.91 -> that there is no time left to experiment and innovate.

21.21 -> Moreover, we need to be thoughtful about where

23.64 -> and how invest our money.

25.65 -> So how can we be more agile, innovate faster,

29.37 -> reducing the risk, and get ready for the next revolution?

34.17 -> My name is Simone D'Antone and I'm business developer leader

37.5 -> for broadcast in AWS and today I will walk you through

41.22 -> the work that review

43.5 -> with a lens on how it can be used

45.87 -> in the media and entertainment industry.

49.38 -> Then you'll hear a story from Sky about not only

52.29 -> how they solve those problem but most important,

54.87 -> how they took advantage of the situation

57.27 -> to reinvent their business.

59.49 -> And at the end there will be time for a deep dive session

63.06 -> on some specific topic.

69 -> I'm pretty sure you're already aware

70.95 -> of what the well-architected framework is,

73.26 -> but for who is not familiar, the well-architected

75.96 -> is basically a collection of guidelines,

78.45 -> design principles and architectural best practices

81.69 -> for designing and running workload in the cloud.

85.56 -> The world architected do not belong

87.21 -> to any specific vertical.

88.86 -> Actually it can be used across any workloads

91.86 -> on any industries.

95.37 -> I will not dive deep each section explain the principles

98.28 -> within, but I'd rather prefer to give you some example

101.25 -> about how it can be adopted on some specific M&E use cases.

106.98 -> As you can see the well-architected

108.18 -> consists of six main pillars,

110.22 -> the first of which is operational excellence.

115.53 -> One of the mantra of the broadcasting industry

117.96 -> I heard several times is if it's working,

121.23 -> do not touch especially during the weekend.

125.43 -> That's because making changes was difficult and risky

128.7 -> and we tend to avoid to touch a production system

131.79 -> and push changes into large releases once in a while.

135.99 -> But in the cloud, constraint of a traditional environment

138.36 -> are removed and you can use the design principles

141.18 -> of the operation excellence pillar to run

143.55 -> your infrastructure as a code and make all changes as a code

147 -> till the point when you can run infrastructure,

149.82 -> use it and destroy when it's no longer needed.

153.15 -> Try to think about production gallery for any given sport.

157.02 -> Not only you can spin up your environment,

159 -> operate it and spin down when it's no longer needed,

162.21 -> but you can also create different templates

164.55 -> depending on the complexity of the event itself.

167.28 -> So for example, you can have a small event

169.95 -> with just four cameras, small mixer and no graphics

173.277 -> and another templates with 40 cameras,

175.83 -> crazy graphics and all the bells and whistles you need.

179.49 -> And those templates are not rigid.

181.53 -> They can be refined the next time based on lessons learnt

185.16 -> or last-minute request, simply changing few line codes.

192.33 -> Security is without question the most important pillar

195.06 -> and number one priority for AWS.

197.91 -> But let's be honest, in the broadcast industry

200.61 -> we never really put attention about how to secure access

204.3 -> or signal streams.

206.28 -> Raise the hand, who never used admin password tool again?

211.62 -> Yeah, password one, two, three way more better.

215.13 -> It's the same for a signal link.

216.51 -> It was essentially a coax cable connected point to point.

220.53 -> It's important to realize that we can be more agile now

223.62 -> and that connection is no longer point-point,

226.14 -> but it is multi-point and multi-axis.

228.96 -> Therefore, not only we should protect the signal in transit

232.41 -> every single time, but we should also think about who

235.68 -> and how can consume that stream.

238.86 -> It's actually very easy and what we've seen is that

241.83 -> once adopted, it become muscle memory till the point

245.34 -> that every engineer, architect, or even manager

248.88 -> is now paying attention about how to secure the streams

251.97 -> or the file as a natural process.

256.65 -> Reliability is by far my most favorite pillar.

260.16 -> We are used to design fully redundant solution,

263.31 -> usually replicating our infrastructure twice

265.83 -> with two complete an independent chains, main and backup.

270.15 -> And on top of that a disaster recovery somewhere

273.18 -> ready to go on air.

275.13 -> And that's because as I love to say,

276.9 -> one frame lost is just lost and every frames matter.

281.91 -> The concept of (indistinct) meeting a frame

284.31 -> during a live event, like if it was an IP packet,

287.61 -> doesn't apply for broadcast.

289.26 -> Every frame matters.

291.93 -> And a frame, depending on the standard,

294.96 -> is essentially alive for 16 milliseconds.

298.59 -> So it's clear that our infrastructure must be designed

301.74 -> accordingly and when, not if, when a problem happen

306.57 -> it must react quickly with no frame loss.

310.65 -> In fact, if there is something we know about the technology

313.32 -> is that it fails.

314.64 -> But now we can unlock new way to manage the failure

318.66 -> with faster and more efficient recovery mechanism

321.57 -> such as cause engineering techniques.

324.66 -> Also we can have a better granular approach

328.35 -> to different business needs.

330.06 -> This is linked to the next pillar, cost optimization.

335.25 -> Think about a linear channel.

337.11 -> If you're a broadcaster, most likely your app

339.66 -> is made by premium channel, content channel

342.93 -> with just movies and TV series and some tier two

346.35 -> and tier three channels.

348.81 -> Let's take for example the kids channel,

351.42 -> which by the way I consider to be mission critical

354 -> because during the day we want our little devils

356.55 -> to stay quiet for a while.

358.83 -> But in the night when the QT monsters are supposed

361.41 -> to be asleep, do we really need main backup,

364.71 -> DR strategy configured, active-active all the time.

368.76 -> Maybe for some of this channel we can turn off the DR

371.73 -> and the backup and keep a cost by infrastructure

375.36 -> ready to go on air when a problem happens.

378.51 -> Your asset is still protected.

380.31 -> It just take a couple of minutes to go back on air.

382.47 -> But perhaps it fits the SLA for that kind of channel.

387.63 -> Plus you can increase your resiliency

390.12 -> with no additional cost.

391.53 -> Having infinite replica in close stand by

394.08 -> ready to be spun up without paying

395.91 -> for an active replica which is doing nothing.

401.49 -> But why do we keep talking about main backup, DR

405.06 -> if we can have an infrastructure serverless and distributed?

409.17 -> What do we mean by serverless?

411.75 -> Let's say you need a way to manage your media assets

415.65 -> and most likely you need a database.

418.68 -> It doesn't mean you need to spend time to install,

421.08 -> maintain and patch the DB engine.

423.39 -> What you need is populate entries in a meaningful way.

427.53 -> That's what servers means.

429 -> You can stay focused on the application layer

431.07 -> while the infrastructure underneath is totally managed

434.13 -> on your behalf.

436.29 -> In distributed,

438.15 -> think about the previous example of the production gallery

441.3 -> but this time imagine that each component,

443.79 -> it's based on node independent from each other.

446.97 -> So you can have one component for the mixer,

449.61 -> another one for the graphics, another one for the audio.

452.31 -> And you can connect those pieces like Lego bricks in minutes

456.42 -> staying focused on the workflow,

458.43 -> creating your custom pipeline without taking care

461.37 -> of the infrastructure.

463.38 -> On top of that, you can now experiment more often

466.47 -> because it's very simple to replace a single component

469.41 -> without affecting the entire chain.

475.56 -> Last year we introduced the new pillar about sustainability

478.74 -> and Amazon as a whole is committed to be powered 100%

482.88 -> by renewable energy by 2025.

486.57 -> I think the broadcast industry can do a lot in this regard.

490.02 -> The most obvious example is about a remote production.

493.65 -> Think about a traditional OB1, a truck

496.35 -> sent on the location of the event to produce

498.75 -> I think about all the crew that must travel

500.91 -> in the same location as well.

503.22 -> Remote production is now real.

506.1 -> and lots of venue are connected.

507.54 -> So what you can do is just send those more crew on site,

510.36 -> let's say cameramen and some engineers,

513.24 -> but you can shift the entire production in the cloud

516.27 -> keeping all the other operators such as director, mixer,

520.2 -> commentary, graphics in their houses

523.98 -> without compromise the production quality.

526.53 -> They can still work like they were sitting close each other,

529.65 -> optimizing their time as they now can produce more event

533.37 -> back to back without wasting time traveling around.

536.91 -> We also a huge direct impact on scope one emission

540.48 -> on the environment.

545.01 -> That was a quick overview about how the well-architected

547.62 -> framework can help media company like yours

550.5 -> to innovate faster.

552.27 -> But now let's listen

555.27 -> what Sky have done and how they implemented

558.09 -> those principles to solve the real life problem

560.64 -> and to evolve and transform their business.

563.97 -> Because sometimes problems can be opportunities.

569.04 -> Please join me welcome Sky's group director of content,

571.83 -> broadcast and platform engineer, Mr. Dave Travis.

575.624 -> (audience applaud)

581.94 -> - Hello everyone,

583.68 -> good to see you.

586.2 -> I'm the guy that has probably one of the longest job titles

589.53 -> in the industry and clearly I need to work

592.14 -> on trying to make that a bit more efficient.

594.12 -> So I'm gonna take you on a bit of a journey,

598.53 -> walk you through where we started, where we are going,

602.73 -> talk to you about what we've done previously

606.12 -> and then what's really changing

608.82 -> the way we work and innovating.

610.92 -> So I'll talk about our journey so far,

614.55 -> I'll talk about our challenges,

616.02 -> I'll talk about our successes

618.42 -> and things that we learned along the way.

623.94 -> Three years ago we established a part of Sky called CT&I,

628.38 -> Content, Technology and Innovation,

631.17 -> and we are a operations and engineering organization

635.13 -> that facilitates workloads for our internal stakeholders.

639.84 -> Sky News, Sky Sports, Sky Creative,

643.89 -> our global OTT organization

647.61 -> and we have the end to end.

649.89 -> So we are a horizontal organization.

652.77 -> We came together as group.

654.66 -> So that combined UK, Germany and Italy,

658.14 -> which you'll hear about is how we've evolved.

661.92 -> So we are about 2,700 people today

665.82 -> and we have an engineering organization of about 650 people.

670.47 -> So I'm gonna give you a little sizzle reel

674.67 -> of what CT&I is all about.

677.942 -> - [Narrator] CT&I, we are content.

681.922 -> (audience applaud)

682.92 -> Our vision is to make your TV and all platforms

685.74 -> come alive like never before.

688.48 -> ♪ Anytime, anyplace, anywhere, anywhere ♪

692.398 -> (upbeat music)

694.05 -> We're innovative and work together hand in hand

697.98 -> with over 2000 channels and content partners.

701.988 -> We are behind the screen for a unique live experience

706.08 -> and provide over 300,000 video on demand hours.

709.884 -> (upbeat music)

712.26 -> We are hands on

714.42 -> and always ready to create whatever needed

717 -> for our platform.

719.01 -> We are creative.

720.315 -> (upbeat music)

721.8 -> We are reliable and cover 99.995% uptime.

727.478 -> (upbeat music)

728.76 -> We are unforgettable moments.

731.711 -> ♪ Call me ♪

732.669 -> (upbeat music)

734.627 -> ♪ Call me, call me any anytime ♪

738.602 -> ♪ Call me ♪

740.31 -> ♪ You know you can call me ♪

743.218 -> ♪ Any day or anytime, call me ♪

746.177 -> CT&I, we are content.

752.91 -> - So wind back the clock three years ago,

756.57 -> we were an organization that had

759.57 -> three different like geographic locations,

762.03 -> we had different engineering teams,

763.32 -> different operational teams

765.72 -> doing pretty much the same thing but very differently.

769.5 -> So we saw an opportunity to come together as a group

772.86 -> organization and try and go on a journey of convergence.

776.73 -> And that journey of convergence, well if you're starting

780.18 -> something new, where should you be building your workloads?

784.59 -> And we decided that really our focus

786.54 -> should be moving to the cloud.

790.41 -> So we first started by setting our operating model

795.96 -> and that was all around aligning our organization.

799.11 -> Getting the structure right, understanding our KPIs,

803.22 -> getting organizational symmetry,

805.11 -> having the right diversity within your organization.

808.74 -> So having shared teams and shared groups,

812.01 -> that's not an easy thing to do.

813.48 -> But once we got that alignment we could then build on that

817.05 -> to go on the next step and the next step.

819.45 -> So then we focused on the transformation piece

823.8 -> and that's the bit we are kind of working on right now

826.53 -> and I'm gonna talk about in a bit more detail.

829.23 -> And then the third part of our pillar

832.26 -> is working on full abstraction.

834.54 -> So today a lot of our organization as you know

838.41 -> in engineering is you have the hardware

841.14 -> and you have the people and they're very much co-located.

844.86 -> And what we really want to do is have a organization

848.55 -> that means that they could operate it from anywhere

850.32 -> in the world.

851.31 -> They could cooperate it from home

852.36 -> and as we saw from COVID,

856.47 -> the life has changed.

858.15 -> COVID was an incredible challenging time for us,

863.55 -> but it was actually probably one of the best things

865.86 -> that happened to us because we did things in six months

869.46 -> that would typically would take us three years.

872.13 -> Take remote production.

873.99 -> We had a three-year vision to to do remote production,

878.28 -> we did it in six months.

879.96 -> We were doing premier league football,

882.12 -> we were having 40 feeds

883.53 -> coming in from every football stadium.

885.57 -> We do cricket with 60 feeds

887.55 -> and so we took one of our studio floors,

890.52 -> a bit of of space like this and we crammed

892.8 -> and we built load of galleries overnight.

895.11 -> So it was a really interesting experience for us.

898.38 -> But that showed the after the possible what you can do

900.51 -> when you bring everybody together and set them

902.7 -> on an objective to go on the journey.

908.01 -> This was our technology environment.

910.05 -> So multiple environments in every country.

914.82 -> We had astronomically long lead times.

918.03 -> So launch a channel in Sky, it's a complex organization.

923.79 -> We were talking somewhere between six to 10 months of time

928.23 -> and it would probably take 60 people as involvements

931.02 -> to make that happen.

933.12 -> We were incredibly resource-hungry.

938.31 -> We had all of these different locations and no consistency,

942.78 -> no ability to scale,

944.97 -> our costs were just through the roof,

947.28 -> we had an aging estate.

950.76 -> Sky's been a very incredibly successful organization,

954.84 -> but it's organically grown.

957.06 -> And so we were hitting like a critical refresh stage

961.98 -> and we were burning carbon like it was going out fashion.

965.85 -> We were sending people all over the world doing OB trucks,

969.96 -> driving these big vehicles around,

972.27 -> then we were losing them for weeks at a time

974.16 -> because they're in Singapore.

980.01 -> And then we just had no ability to scale at all.

983.34 -> So everything was just slow, mundane.

986.43 -> I think our stakeholders looked at us like,

989.04 -> oh, it's that slow organization.

991.71 -> So we decided to set a vision of where we wanted to go

995.16 -> and take our people on that journey

997.47 -> and that this is what we were targeting.

1000.02 -> The north star for us was to automate and rapidly deploy

1004.76 -> services at speed.

1006.92 -> So go from 10 months

1012.32 -> and I'll talk about how we are deploying in four minutes.

1015.65 -> So that in its own right is a massive efficiency gain

1020.27 -> and our stakeholders really

1023.21 -> then appreciated that we can launch source services

1025.97 -> and we can innovate quicker.

1028.82 -> We centralized all of our capabilities

1032.24 -> and looked at how we can abstract all of the functions,

1036.65 -> storage databases, data,

1039.71 -> having data mesh, data mobility,

1043.94 -> having the ability to have a consolidated

1046.97 -> operational footprint,

1050.21 -> offer the business a lower total cost of ownership.

1053.81 -> And that's quite critical really when you go on the journey

1056.81 -> and you're trying to sell it to the finance organization

1060.14 -> who just see a load of opex and we'll cover a bit more

1063.32 -> about that in a little while.

1066.47 -> And working for an organization like Comcast,

1070.995 -> we don't get audited, we get continuously audited.

1073.73 -> It's like nonstop.

1078.71 -> If I look in a lot of our environments,

1080.48 -> our playout environment had like a 100 security failures

1085.22 -> and vulnerabilities identified.

1087.59 -> And back then, you know, it's that oh, we can't touch it.

1090.56 -> A little bit of what Simone was saying earlier,

1092.21 -> you know, we have to leave it alone, it's precious.

1096.41 -> But equally the organization didn't wanna take the risk

1099.89 -> on security so managing that was challenging.

1103.31 -> Targeting an ultra secure model was part of the vision.

1107.36 -> Reducing our carbon footprint

1110.57 -> and then having the ability

1112.34 -> to scale up and down resources on demand.

1115.52 -> So that was the vision of what we wanted to do.

1120.2 -> And within our ecosystem, this is kind of like

1124.34 -> the stack in my area.

1127.79 -> We decided, we initially started doing things in the cloud

1134.268 -> but when before I joined there was a little bit of a,

1138.95 -> we were doing things but we weren't really

1142.67 -> doing it correctly.

1143.9 -> We were rushing at it, we were going too quick.

1146.03 -> We would launch a service in the cloud

1147.5 -> and have no way of managing it.

1149.9 -> So we then took a big step back and said

1153.32 -> we need to focus on the foundational components first

1156.05 -> and get that absolutely right.

1157.4 -> So we need to focus around our deployment,

1159.77 -> what's our deployment strategy?

1161.54 -> How will we manage that sort of ecosystem

1165.65 -> so we can continuously deploy at speed and we can do

1168.527 -> all the security patching and everything on a daily basis.

1172.4 -> We wanted to have a storage strategy.

1175.73 -> So we wanted to have a federated

1177.59 -> global name space that could

1181.76 -> talk to systems in the cloud S3 Glacier

1184.88 -> but equally talk to all of our systems on-prem.

1189.59 -> And then have a data strategy that could take us

1192.92 -> on that journey.

1193.79 -> So once we built these foundations,

1196.4 -> we could integrate our teams to provide them full autonomy

1200.57 -> to build these environments that you see here.

1203.12 -> Now, not all of them are suitable for the cloud.

1206.21 -> You can't take a studio physical facility

1209.33 -> and put that in the cloud.

1211.01 -> But you can take all of the hardware,

1213.8 -> the traditional hardware and strive for software.

1217.7 -> And through perseverance we really pushed

1221.81 -> and had that desire to find and select vendors

1225.02 -> that could go on the journey with us and partner.

1227.9 -> And so we now have a lot of our post-production and editing

1233.18 -> running in the cloud.

1235.76 -> We have cloud production, I'm gonna talk about that,

1239.18 -> it's a case study I'm gonna go through.

1241.13 -> But we have a cloud production environment

1242.93 -> which can do live sports and live news.

1246.59 -> We have 120 channels

1250.19 -> running in playout in the cloud.

1253.34 -> We are doing video encoding, stack boxing in the cloud.

1258.232 -> And so we are starting to see an evolution where

1261.89 -> 50% of our workloads are now in the cloud and our target

1266.12 -> is to try and get to a 100% over the next five years.

1270.62 -> So Sky also has a big commitment to sustainability.

1276.11 -> We want to be carbon neutral by 2030.

1279.98 -> So everything that we do with the cloud

1283.04 -> has really helped and enabled us to go on that journey.

1286.97 -> We've done things like we ran our first carbon neutral

1289.97 -> football game in partnership with the Premier League.

1293.6 -> And all of these things are just massively helping

1296.51 -> the environment and helping us reach that target.

1300.86 -> So this is the journey that we've been on

1304.31 -> and how it started.

1306.53 -> We had a fire to one of our facilities in Rome in 2019.

1312.41 -> The facility burnt to the entire ground, it was gone.

1318.05 -> We have had COVID, we have had Brexit,

1322.67 -> we had the Queen pass away.

1325.52 -> It's been an interesting environment to work

1329.42 -> and throughout all of this unusual organization

1331.94 -> has to continue and our sports have to continue

1334.88 -> and we have to carry on pleasing our customers.

1337.55 -> So we started with the business transformation,

1341.36 -> we established the organization,

1343.88 -> we bought the organization together and converged it.

1348.14 -> We set the model of what we wanted to achieve and the vision

1351.8 -> of what you are working towards, brought the alignment

1353.9 -> and everybody to work towards that common goal.

1357.29 -> And then we focus on how we could develop

1359.9 -> and improve the way we work,

1362.6 -> which then enabled us to get the engineering components

1365.42 -> right and and set clear engineering principles.

1369.98 -> What is it we are looking for from our vendors?

1373.01 -> Do we want them to be cloud-native?

1375.2 -> 'Cause I tell you what, a lot of them

1376.79 -> are all monolithic lift and shift.

1378.98 -> That's not what you want.

1380.57 -> You wanna be working in a very much pure software space.

1385.34 -> So everything in what we were trying to achieve

1388.13 -> is we would accept feature gaps to ensure

1392.27 -> that we've got the right deployment architecture

1395.3 -> and that has massively benefited us

1397.58 -> with what I'll talk about in a minute.

1400.22 -> So, so far we have

1404.84 -> our content supply chain is fully in the cloud.

1407.87 -> We have 500 content partners,

1411.95 -> we have hundreds of petabytes of storage.

1415.04 -> We manage, I think it's like 250,000 assets a year.

1419.12 -> And so that's all running.

1420.41 -> We partner with, our good partners are in the room SDVI,

1424.52 -> and that's been incredibly successful

1427.28 -> and that has then led us to then look at other parts

1430.01 -> of our ecosystem.

1431.78 -> And as you can see here at the bottom,

1434.69 -> we started doing cloud production.

1437.15 -> Cloud production started off with some lower tier sports

1442.4 -> and then we had the opportunity to do COP26 last year.

1447.23 -> And the news organization came to us and said,

1450.02 -> we wanna have a dedicated channel.

1451.76 -> It's in Scotland, Sky are one of the main sponsors.

1454.88 -> How can we get behind this?

1457.4 -> Typically that would've been 10 months,

1462.11 -> few million pounds, we'll get you on air.

1465.26 -> But we didn't have that time.

1467.27 -> So we used our cloud production environment,

1469.25 -> we deployed it all of the system,

1471.98 -> fully remote production in the cloud in weeks

1476.21 -> and saved, we had a factor of 10 carbon reduction

1480.38 -> in fossils and the entire

1484.76 -> economics was like just dropped off a cliff.

1487.1 -> It was just like ta-da, here you're gonna have one of these.

1490.67 -> And then we are currently in the process of building

1494.87 -> a new production content management system

1497.27 -> in partnership with VD Spine that will host

1501.05 -> all our production for all our sports.

1504.02 -> And then in long term we have looked to try and take that

1506.57 -> for our newsroom systems.

1509.51 -> We have just successfully migrated the 120 channels

1513.29 -> I was talking about.

1515.63 -> And we launch our first stat muck in the cloud in January.

1520.49 -> So it's been an incredible journey

1522.95 -> and we're really looking forward

1524.12 -> to where we're going the future.

1526.94 -> But it all comes with a lot of challenges

1530.51 -> and I think these challenges are common

1532.64 -> to probably most people in the room.

1536.06 -> There is not a talent shortage in the UK or it's in Germany.

1540.83 -> There is a global talent shortage,

1543.59 -> which is incredibly difficult to address.

1547.1 -> So we have to take our workforce and re-skill them

1551.63 -> and go on that journey, which takes time

1554.582 -> and it's all part of that big consideration

1558.23 -> with our transformation journey.

1560.75 -> We have a lack of capacity.

1563.66 -> I use the analogy working at Sky is like bombing down

1566.45 -> the motorway 100 miles an hour while changing the paint

1568.61 -> on the car and the wheels and everything.

1570.53 -> But on top of that you gotta transform.

1572.75 -> So that's a real challenge,

1576.23 -> especially if our external stakeholders.

1578.66 -> They could take my entire team

1580.76 -> and fill up my roadmap exclusively just on the old tech.

1585.02 -> So that's a real problem for us.

1589.82 -> The culture of what you're trying to do.

1593.54 -> Broadcast engineers

1595.73 -> like to sit down, assess,

1598.22 -> take their time, very calculated

1601.85 -> and that fundamentally

1604.67 -> meant that the speed you move at is quite slow.

1608.09 -> So we have to get to this point of we wanna learn fast,

1612.59 -> we wanna fail fast.

1613.73 -> So the more we fail, the more we can learn,

1616.01 -> the more we adapt and pivot and go onto that journey.

1620.48 -> We had, and still do, a vast amount of tech debt but

1627.95 -> we see 75% of our tech debt will be eroded

1632.45 -> by the transformation we're doing into the cloud.

1634.25 -> So step by step that's gonna be eroded and it means that

1638.3 -> we don't have to invest large sums of CapEx in hardware

1641.09 -> which I have for five years and then have to throw away

1643.16 -> and install over and over and over.

1645.56 -> That will just go away

1646.55 -> because we'll be in a continuous environment.

1650.42 -> Cloud native, I said it before,

1653.96 -> vendors in our industry just are struggling to change

1658.82 -> and that's because they have the challenges

1662.45 -> of their hardware business

1663.95 -> and they don't have enough funding to invest

1665.72 -> in the software business.

1667.37 -> And here I see endless evidence

1669.92 -> of how this is challenging for them.

1672.59 -> So you have to identify the right partners and some of them

1676.37 -> may even be startups and take a risk

1679.22 -> because they're better than the traditional vendors

1680.817 -> that have typically been out there.

1683.93 -> And then fin ops is another challenge because

1690.5 -> when you start doing millions and millions and millions

1692.99 -> of dollars in the cloud,

1694.91 -> typically the finance team come along and start asking you

1697.31 -> why you're doing it and what you're doing it

1698.57 -> and how can you do it cheaper.

1700.04 -> So you have to get right behind that to have the right level

1705.29 -> of intelligence to manage those workloads

1709.28 -> and provide educated information back

1711.92 -> to those commercial parts of the business.

1716.21 -> So I'm gonna get into a few case studies.

1721.04 -> So this is our cloud production environment.

1724.85 -> We are working

1727.37 -> on a sort of a take the lower value tier sports

1731.57 -> we have today, the lower value rights,

1734.51 -> take them on the journey into the cloud.

1736.46 -> And then as we learn and adapt our target is is that

1740.06 -> we can do our highest valuable assets like Premier League,

1743.72 -> Champions League, NFL, all in the cloud ultimately.

1749.15 -> We're not there yet because audio is an utter nightmare.

1753.95 -> There is not one single solution out there

1756.14 -> that we can really identify

1759.38 -> to do audio at the scale that we need it.

1762.65 -> But you know, we are getting there slowly and slowly.

1765.05 -> So we have partnered and created an ecosystem.

1769.43 -> We at Sky

1771.71 -> like to have best of breed technology

1774.29 -> and we also want to have in our live environment

1778.19 -> like what we have in the content supply chain.

1782.12 -> You wanna have a federated approach so you can interchange

1785.27 -> products very quickly as new things come on.

1788.12 -> So we are working with people like Vis,

1790.58 -> we work with LiveView, now it sounds a bit weird,

1793.52 -> LiveView, why you working with LiveView?

1794.81 -> Well, we decided that actually we can break apart

1798.47 -> how we do replay and how we do live because those two

1802.79 -> don't need to be synchronized with each other.

1805.19 -> So we have stressed and tested and tried different models

1809.78 -> to try and get the right architecture.

1812.3 -> The dream for us is to go camera head to the cloud,

1817.61 -> everything there.

1818.81 -> But again, connectivity's not sufficient in all locations.

1823.85 -> Formats are very inconsistent

1829.22 -> and again, some of the technology

1831.14 -> in these locations are restricted.

1835.46 -> So this is where we're at today

1840.98 -> and this is what we produce so far.

1843.151 -> So we got women's netball running on the platform.

1847.25 -> We've been doing some women's soccer on our platform

1851.96 -> and as I was talking about earlier,

1854.96 -> what we did with COP26 was just like monumental moment

1859.13 -> for us where we realized

1862.13 -> this actually is exactly the same as the newsroom

1864.71 -> we have on the ground.

1866.48 -> And there is no difference of what we had.

1870.05 -> So we are going through the journey

1873.47 -> of doing more and more and more

1875.84 -> and then we started doing red button services.

1877.88 -> So our interactive services we have today,

1880.67 -> we have the ability to have different holes for a golf

1883.22 -> or you can have a different, you can switch to onboard.

1886.22 -> That was just other galleries we had

1889.34 -> that were doing exactly the same.

1891.68 -> We took the decision to move all of those workflows

1894.08 -> into the cloud because it's not the premium output,

1899.12 -> it's your secondary output.

1901.07 -> And that's been incredibly successful and proven that

1904.04 -> again, we can go further with what we're doing in the cloud.

1911.15 -> So then we went on to, we've been on the journey of playout.

1917.33 -> And playout, a lot of people have been doing this right

1920.69 -> and it's evolving but we had a huge amount of end of life.

1926.39 -> I was quite surprised when I joined the organization that

1929.36 -> we were like two generations behind in terms of playout.

1932.99 -> So we wanted do a massive generation jump

1936.322 -> to something new and we had playout,

1941.42 -> I think it was on six different locations.

1944.3 -> So again there was a wonderful opportunity

1947.21 -> to bring all of those teams together and

1952.07 -> converge and consolidate.

1954.86 -> So we realized though that we don't have to have,

1959.21 -> not one size fits all.

1961.82 -> And we decided that the 200 channels

1965.6 -> we have in our portfolio of ones that we produce ourselves,

1970.52 -> we came up this model of a content TV and event TV.

1974 -> So content TV is quite lightweight,

1977.36 -> it's mostly playing out file-based content.

1980.39 -> It has some live, not much,

1984.68 -> and we can have an ops light model.

1988.19 -> So all of those services that you see here

1990.8 -> now running in the cloud and it's been game changing for us.

1995.72 -> We've now gone from six sites to having two teams

1999.05 -> that are load balanced in two markets and all the tech

2004 -> is just virtual now, so that's a really big step.

2007.24 -> We're then working on event TV now which is how we do

2012.16 -> very sophisticated, very interactive playout

2016.99 -> where we are working with the news team continuously

2019.99 -> or we're working with the galleries

2021.82 -> and on the production side.

2024.88 -> So that's what you'd see here for Sky Sports and Sky News.

2029.32 -> That's been challenging because again,

2032.5 -> most of the vendors out in the industry are not giving us

2036.55 -> what we require, they don't meet our engineering principles.

2041.05 -> So we are now reevaluating, we went on a journey,

2043.87 -> we started, we moved all of our disaster recovery for

2049.06 -> Italy into the cloud.

2050.787 -> 'Cause we had to outsource it

2052.36 -> 'cause the facility burned down.

2054.01 -> We have migrated it with a partner now

2056.71 -> but we've stopped and said

2058.54 -> we don't think we can continue on that journey.

2060.34 -> So we are now reevaluating the software provider there

2064.75 -> and we are either looking at partnering with one

2068.98 -> or two other organizations or we may even well develop

2072.13 -> with some gaming technology to take us on that journey.

2078.07 -> So this is what the architecture looks

2079.9 -> for our playout environment.

2082.18 -> Everything we do is active-active.

2085.36 -> The partner that we have that we use

2086.98 -> for content TV is BC Next

2089.38 -> which is a small company.

2093.13 -> Nine people I think are in the organization.

2096.88 -> They are the best partner I've ever worked with.

2099.46 -> They deploy at speed, they get the industry,

2103.57 -> they understand what we were working towards.

2107.74 -> And so now we are doing,

2111.593 -> I think Comcast was saying, we had a Comcast audit,

2114.55 -> like you like the example now of what you should be doing

2117.73 -> from a security perspective.

2119.59 -> 'Cause all our pipelines and the way we deploy

2121.54 -> as security by design

2123.73 -> and we are just doing regular daily updates and

2129.949 -> it's good, we don't have 24/7 engineering teams anymore.

2133.78 -> If something breaks the guys, it just carries on running.

2136.45 -> They'll come in on on Monday and fix it.

2138.73 -> And so it's been a real like game changer and a shift

2142.66 -> to the way we execute and manage our workloads.

2150.1 -> And in the final case study

2152.38 -> that I wanted to share with you today,

2154.06 -> and please feel free if anyone's got questions

2156.25 -> about, just grab me afterwards,

2157.6 -> I'm happy to talk in more detail but,

2161.2 -> we as an organization

2164.71 -> have free DTH

2168.098 -> director home satellite platforms,

2172.54 -> which was predominantly,

2175.51 -> well, it's all hardware based, loads of data centers,

2178.72 -> loads of racks of encoders and stat marks and things.

2182.56 -> We would again distribute it over multiple sites.

2186.79 -> And we got six uplink facilities, et cetera, et cetera.

2191.59 -> So again, similar problems that you see

2194.23 -> I've shared from other slides.

2196.81 -> And we then decided that we really want to build

2200.95 -> a software architecture for encoding video.

2204.46 -> So this is what we have built

2209.5 -> and this is the ability to take

2213.13 -> content from on-prem into the cloud

2218.56 -> or take content from our cloud playout platform

2222.19 -> into the software defining coding platform

2225.43 -> or from our cloud production entity.

2227.86 -> You start seeing how our ecosystem then all starts going

2230.65 -> cloud, cloud, cloud and we can get the then

2233.74 -> way more efficiencies.

2235.69 -> So we've built, again, active-active architecture.

2241 -> In this space, we work with,

2244.99 -> we selected three vendors

2246.58 -> that meet our engineering principle

2249.22 -> and focused on building an orchestration layer

2252.43 -> that allows us to interchange the products underneath.

2255.58 -> Which then means suddenly our software defining coding

2258.82 -> platform becomes like a content supply platform.

2261.82 -> Today we work with Elemental

2264.67 -> but we also work with Telestream

2266.26 -> and those guys in the content supply.

2269.71 -> Well then we can start doing the same thing

2272.05 -> on our video encoding technology.

2273.55 -> So we can interchange products and change and flex

2277 -> so we can be really quick and agile,

2279.34 -> which has been phenomenal game changer

2282.07 -> for the way we run our organization.

2287.65 -> So this is the roadmap for next year in terms of migration.

2294.43 -> Next month we'll be launching some German services

2298.9 -> in the cloud.

2300.73 -> We then through all of next year,

2303.55 -> we'll be migrating all of the services you see listed.

2308.35 -> So then we'll have hundreds of channels,

2311.17 -> all running workloads in the cloud,

2313.6 -> the full flexibility, the benefits of reliability,

2317.44 -> which really sort of honed into what

2320.98 -> Simone shared with us earlier around the AWS framework.

2327.1 -> So it's been,

2329.11 -> I think this really summarizes where we got overall.

2335.08 -> We've just consolidated loads of facilities

2338.65 -> into sort of an A and B operation.

2342.58 -> We've had huge amounts of gains with the way we deploy

2346.72 -> and launch services.

2349.99 -> We've seen a massive reduction in our carbon footprint.

2354.01 -> And that's predominantly because,

2357.13 -> not because the cloud has a magic wand that allows you to,

2364.15 -> because it's ultimately hardware just somewhere else, right?

2366.76 -> But it's because we don't have to build

2369.49 -> the lowest common denominator.

2371.26 -> We can literally for cloud, you know,

2374.44 -> for one sport we only need this, this and this.

2377.44 -> But then for another sport it may be 20 different services

2382.3 -> but we only use it for like the hour, the two hour,

2384.94 -> the three hour and then we turn it off.

2389.19 -> In our on-prem facilities, we've got thousands of racks

2392.65 -> of equipment which pretty much is only used for like

2396.01 -> two or three times a week for like a few hours.

2399.52 -> But the rest of the time we never turn it off

2401.02 -> 'cause you don't wanna touch it,

2401.89 -> you won't wanna leave it alone because it will break.

2405.04 -> We don't have that anymore.

2406.57 -> So that's where the efficiencies gains come.

2411.61 -> We've improved our quality because we now have consistency

2414.82 -> the way we architect.

2417.4 -> Everything's active-active and we can do all the relevant

2421.3 -> movement of workloads and things to keep us flexible.

2425.41 -> We now have

2427.27 -> very mature security posture

2430.78 -> doing continuous patching and things like that.

2433.57 -> And we can then we are looking at, we use preventative

2437.74 -> maintenance so we can analyze data to understand

2441.16 -> things will go wrong before they go wrong.

2444.28 -> And again, it's been an incredible journey.

2449.382 -> I think somebody asked me yesterday when,

2452.32 -> actually no, somebody asked me this morning,

2454.48 -> when do you think you'll be at full saturation?

2457.12 -> We got a long way to go.

2460.24 -> But we think that it's incredibly exciting.

2463.57 -> We're very proud of what our teams have done

2467.26 -> and the talent and the way they got behind this.

2471.85 -> So I think that was the end of just an overview

2476.65 -> of what we've done and Simone's gonna come up

2479.86 -> and I think we're gonna have a further chat.

2483.547 -> (audience applaud)

2492.94 -> - I'm here to give you a hard time.

2496.51 -> - I hope not.

2498.457 -> - So thank you very much, it's always a pleasure.

2500.89 -> I mean, being exposed to such a inspiring vision

2504.07 -> and I hope the audience is happy as much as I am.

2507.61 -> I took the liberty to write down some question

2509.86 -> which I hope reflects the audience vote as well.

2513.01 -> And the first will be about the principle

2516.52 -> of the transformation.

2517.45 -> You mentioned about leadership principles

2520.42 -> and we know that in order to create

2522.1 -> this huge transformation,

2524.08 -> you need to set your north star.

2525.88 -> Can you tell us a little bit more about how and where

2529.3 -> are these principles? - Yeah.

2533.29 -> The principles always are evolving

2536.11 -> but we've wanted to foster a culture of innovation

2539.38 -> and oh, yeah, we're sharing it here actually,

2542.74 -> is really focusing on those foundational things always

2547.03 -> and making sure that the vendors

2549.637 -> and the people we partner with are reliable

2554.38 -> and flexible and meet these principles

2557.5 -> so that we can get the gains.

2559.09 -> And, you know, everybody can probably read the slides

2562.45 -> but, you know, reliability, scalability

2565.42 -> are incredibly important, the security by design,

2570.43 -> having the ability to share and leverage resources

2573.43 -> across the organization

2576.22 -> and really channeling in reducing complexity.

2579.97 -> Because quite often, we'll I think sometimes

2583.84 -> try and overcomplicate what we do.

2587.2 -> So taking a step back and reimagine the way you manage

2590.86 -> those workloads to try and work round

2593.86 -> 'cause there are lots of challenges with the transformation.

2597.37 -> You know, like the example I gave around replay,

2601.87 -> we were trying to synchronize all the cameras

2604.33 -> on remote production

2605.53 -> but then we realized well actually we don't need to,

2607.57 -> we can break the model apart

2610.84 -> and then focusing on thinking cloud, thinking data,

2614.8 -> think et cetera, et cetera.

2617.47 -> - Those are mainly technical transformation.

2619.36 -> But how did you handle the top down transformation

2623.23 -> and include all the C level into this discussion?

2627.46 -> - It was a challenge.

2630.01 -> It was, I think the thing that I would say is,

2634.03 -> and I'd say to all my, we have to think commercially.

2638.29 -> We've gotta be thinking that we don't just do this

2641.59 -> because it's fun.

2643.75 -> We do it for a reason, we do it for our customers.

2647.08 -> And so we have to demonstrate the value.

2650.83 -> And typically that value usually comes down to cents

2653.23 -> and dollars, right?

2654.79 -> So we focused on building good TCO business cases,

2661.3 -> the demonstrative return.

2663.67 -> Typically those business cases were more like five,

2668.83 -> they were like six to seven years

2671.08 -> because you wanna include your refresh cycle, right?

2674.44 -> And you need to also consider that when you have on-prem

2677.86 -> equipment, you need some engineers there,

2680.41 -> you need to then provide toilets,

2682.42 -> you've got kitchen facilities,

2683.95 -> you must include all of these things

2685.66 -> because it's false economics if you don't.

2690.37 -> So we did a lot of work and actually that was quite

2693.97 -> surprising to most engineers 'cause they just say,

2696.1 -> oh, it's old and we just need to do it over here.

2697.6 -> Well, that's not the answer that the seed level looking for.

2700.36 -> - In talking about economics, feel free to avoid the answer,

2703.15 -> but how would you handle the shift

2706.3 -> from massive CapEx to opex?

2708.58 -> I think there is a lot of work internally also to justify us

2711.37 -> measuring it. - Yeah.

2715.15 -> We are addicted to CapEx.

2718.03 -> I think that's common in a lot of our sector.

2721.48 -> It is challenging.

2723.31 -> So cloud comes with opex,

2726.52 -> you don't have the luxury of being able to depreciate

2728.74 -> the assets and things.

2731.815 -> But we try and offset the opex with efficiencies.

2736.21 -> So if we can make efficiencies in the way we operate,

2739.96 -> then that can make way for us to fund what we do

2742.72 -> in the cloud.

2744.13 -> And then overall from a cash perspective,

2748 -> it should reduce. - It works, makes sense.

2754.54 -> I know that it's always great when, you know,

2758.83 -> plans deliver the expected results,

2760.87 -> but you mentioned that we learn more about our mistakes.

2765.04 -> I can talk hours about my mistakes just for the last week

2767.71 -> but what kind of mistakes have you done

2771.1 -> and what do you learn from these mistakes?

2777.94 -> - In my previous role,

2781.742 -> we were on a similar journey actually,

2783.43 -> and we were very much pushing for software.

2788.26 -> But we decided to go predominantly on-prem

2792.19 -> and private cloud and built our own private cloud.

2796.27 -> That was a horrible lesson to learn that that was not,

2799.69 -> because you end up with all the same challenges.

2801.79 -> Yes you are, maybe it is all cots and it's all commodity,

2806.054 -> but managing capacity, managing black boxes

2809.41 -> is not where we as an engineering organization

2812.89 -> should be channeling our energy.

2815.56 -> We should be channeling our energy into

2819.79 -> adding value to the customer and do discreet offerings

2824.23 -> and things that have never been done before

2825.73 -> rather than managing environments and back boxes.

2828.989 -> And that was a big lesson for me and I think something that

2833.35 -> hopefully I've offered to Sky.

2835.69 -> Maybe my other regret was maybe one too many bottles

2837.7 -> of wine last night, but you know.

2841.3 -> - Fair enough.

2844.63 -> What about

2847.18 -> training and, I mean,

2849.16 -> being confident of what you're doing,

2851.62 -> of course you require training, you require cultural shift,

2854.05 -> you require a lot of

2857.05 -> time and effort to study and innovate yourself.

2860.59 -> How do you think the training helped you in your team?

2863.56 -> - Well, training's incredibly important.

2866.17 -> And as I was saying earlier with the talent shortage,

2868.78 -> if you don't train, you've really got no people

2870.73 -> to do the work and take you on that journey.

2876.55 -> I mean, Amazon have been incredibly supportive

2879.4 -> and provided a lot of training

2880.87 -> and then we've been doing a lot of skills assessment

2883.66 -> looking at what does our engineer look like in the future

2885.94 -> and then how do we get them accredited?

2887.38 -> So we'd build like this,

2889.36 -> that's what a perfect one would look like.

2892.24 -> Then we have this, where are the training gaps?

2894.49 -> Try and identify them and then try and over a three-period

2899.05 -> try and give them the right skills to go on that,

2901.51 -> you know, for the future. - Right.

2904.45 -> You mentioned also about,

2906.13 -> you talked about the partner and some gap in that era.

2908.8 -> So probably training can apply also in the partner.

2912.4 -> Which gap do you see in this industry?

2914.2 -> You clearly measuring the audio.

2915.85 -> Is there something else that we can do

2917.89 -> or we can try to find some solution for you?

2921.97 -> - Well, it's the cloud,

2924.404 -> the lack of vendors that meet our engineering principle

2927.52 -> is always a big one.

2929.92 -> Audio is extremely challenging.

2934.48 -> I think there's still a long way to go with latency

2939.43 -> and, you know, obviously if we can deliver video

2943.36 -> and produce video at lower,

2946.91 -> you know, bit, we can consume less.

2949.87 -> So obviously there should always be a strive for more

2952.66 -> optimization in terms of, you know, really we need

2957.036 -> our vendors to be focusing on serverless

2960.28 -> and things like that.

2961.683 -> - I think that was mentioned before,

2963.43 -> absolutely, makes sense.

2965.35 -> And about sustainability,

2968.23 -> Amazon and Sky have a lot in common.

2970.15 -> We had just seen that COP26 was a good example

2973.21 -> and of course we are working in this direction as well.

2975.82 -> Can you double down a little bit on sustainability

2978.22 -> in general and what's our goal, your vision for the future?

2983.86 -> - Yeah, I mean, probably not much more than probably

2987.7 -> what I've said already, but, you know,

2991.12 -> we've gotta find a way to be more efficient

2996.43 -> and if that's, I think, you know, using

3000.6 -> vendors that are using renewable energy sources

3004.68 -> is incredibly important.

3006.63 -> I think you guys are targeting 24 or five,

3010.59 -> something like that, you've got quite an ambitious.

3013.59 -> So to partnering with people that do focus on sustainability

3016.32 -> in the way they manufacture, the way that they build,

3018.9 -> the way they run.

3022.62 -> Trying not to have, you know,

3024.63 -> lots of people coming into the office,

3026.46 -> you know, everything helps.

3028.17 -> - So let's say all the goals must be aligned

3030.212 -> in term of innovation, in term of products,

3032.64 -> in term of sustainability.

3034.11 -> As long as these goals are aligned, we are fine, right?

3037.65 -> - Well, we hope so, but, yeah. - All right.

3040.41 -> I think we have time for last question and

3043.5 -> that comes from my experience because we talk

3045.87 -> with some friends and that describe my job,

3048.69 -> they think it's all about fun.

3050.34 -> It's, you know, media entertainment, so it's just fun.

3054.359 -> Yeah, I know.

3056.7 -> It's stressful.

3057.608 -> - We're talking about media and entertainment, eh?

3059.514 -> - Yeah, they don't understand the complexity to deliver

3061.47 -> the favorite content in the best possible way.

3065.67 -> But you know what, ultimately I think they're right

3068.55 -> and I think fortunate to be part of this industry,

3071.55 -> it's such a dynamic and rich environment.

3074.37 -> So how much fun is in there from your point of view?

3079.65 -> - Well, the fun comes from the passion, I think.

3082.23 -> And if,

3085.02 -> I think when we set the strategy

3087.63 -> of where we wanted to go,

3090.42 -> I think then empowering the guys to say,

3093.78 -> that's what we want to go and do.

3095.367 -> But you need to fulfill it and find a way

3097.83 -> to, that was good for them.

3100.77 -> They then got very passionate and everybody has good time.

3103.95 -> I mean, they work incredibly hard.

3106.38 -> And I overcome a lot of challenges,

3108.54 -> but the passion that comes through from the team,

3111.66 -> I think creates that fun environment.

3114.707 -> We spend way too long at work, right?

3116.79 -> So if it was all boring

3118.77 -> and not exciting- - There's no reason.

3121.53 -> - I certainly wouldn't wanna come in

3122.55 -> and I don't expect my teams to come in either.

3124.59 -> So, you know, we've gotta embrace and have that passion.

3129.7 -> - And I think that's the perfect farewell message

3132.09 -> to conclude this session.

3133.2 -> So David, thank you very much

3135.33 -> for your time. - Thank you.

3136.604 -> - And thank you everybody for being here.

3137.94 -> Enjoy the rest of re:Invent.

3139.479 -> (audience applaud)

Source: https://www.youtube.com/watch?v=VjbHeRY2Ze4