The sustainable telco network dilemma: Matching net-zero goals with exploding customer needs

0:03 so hi everyone thank you very much for joining today's webinar um on the sustainable telone Network dilemma

0:10 matching that zero goals with exploding customer demand um so today we have

0:15 quite a packed agenda so we'll try and keep the introduction as short as possible um but we'll start once the

0:21 introductions finished with a presentation from Partners um that's backed by an extensive research program

0:27 that has been ongoing for the past few months and then we'll move to Juniper's presentation and armit will take us

0:32 through that and then we'll have time for panel discussion an audience Q&A at the

0:37 end so just a little bit of background on the webinar because I'm sure not all of you are familiar with it um so

0:43 essentially you're all in list and only mode um but if you do have any issues with audio please leave us a comment in

0:49 the questions Tab and one of our team members will assist you we have a Q&A session at the end as mentioned so

0:54 please feel free to submit any questions for our presenters throughout the session um that's also available using

1:00 the questions tab um and everyone who's registered for this session will receive the slides the recording um shortly

1:07 after the session to watch back and to share with colleagues if you are on social media please feel free to tweet

1:13 us um ATL Partners so a little bit of background on our panelists so of course we have

1:19 Phil from STL the managing director of Consulting we have Amit from juniper so vpa product management um we have T from

1:28 telenor um so SVP Network and Cloud technology strategy armed from vone

1:33 Egypt um he's part of the tech center senior solution architect as well as L from light storm and I'm sure not all of

1:39 you have heard of light storm so it's a relatively new operator but they're growing incredibly quickly um they're

1:45 based in India with operations in multiple jurisdictions particularly in Asia um and they built a network with

1:50 cloud in mind both as a customer and as an enabler so essentially they can be quite well described as a data center

1:57 neutral carrier um we do Ur you to check out all the panelists on LinkedIn um

2:02 while we're going through the slides and also after the session um but yeah really looking forward to

2:07 this so if we move on I'll pass over to Phil to take us through asl's

2:13 presentation great so thank you will and welcome everybody good morning good

2:18 afternoon good evening for some of you dialing in from um from Asia um so what

2:24 I'd like to do today is just um set out very briefly um some of the key thoughts

2:30 and findings the situation that we Face ourselves in Industry you're all

2:35 familiar with um you know we are seeing significant and increasingly

2:41 unpredictable growth in traffic so this traffic is is hetrogeneous it's

2:46 unpredictable and it's distributed we're seeing AI um and the growth in Ai and

2:53 iot and then gamings amongst the drivers there so we all know this and we all

2:58 know it's increasingly difficult to predict what and how those patterns are um at the same time most operators

3:06 globally are um at least reporting and many of them setting clear Net Zero

3:11 targets um so they have sustainability commitments and what we finding is that

3:17 um a lot of operators are making good progress on the quick wins as we say but

3:23 as they get Beyond those quick wins it becomes harder partly because the reporting rules are becoming tighter

3:29 what counts as emissions is becoming the definitions are becoming stricter but also um you you you've you've taken the

3:37 the first easier steps and then the steps are getting harder uh when it comes to decarbonizing the network so

3:44 yeah the key question is you've got you you want to avoid the bottlenecks you certainly want to avoid impacting

3:50 customers and having an impact on their quality of service and you you know you need to to drive down your um your

3:58 emissions and obviously maintain margins so how do you overcome this Challenge and that's really what

4:04 we're going to be setting out and talking about today so a little bit of background what do we do as part of this

4:11 assignment we went and spoke to um operators globally um about how they

4:17 were approaching this Challenge and what they saw as the key roots to resolving it so this is the geographic coverage of

4:23 the research um and here are some of the roles of those involved so that's that's a little bit of background about the

4:28 research and now I'm going to just sort of share the key findings of that research and The Wider work that STL

4:35 Partners has been doing over the years around this topic so the broad sort of

4:40 finding is that um you know and this is an oversimplification but the

4:45 traditional economics of networking um are such that as you increase demand or

4:52 as the the the the orange line here on the left demand increases for services

4:58 in line with that so do the result ources required to support that and I I

5:03 guess the challenge we are faced with is how can we move from that world to a

5:08 world which I've described here as hyperscale economics and I think hyperscale economics isn't just about

5:14 hyperscale doesn't mean just big what it means is the the scale um is is

5:21 different so the scale between your resource base and your capability base

5:28 um becomes separated so what you can achieve is either exponential growth in

5:33 capabilities while maintaining a linear um growth in resources or better still

5:40 maintaining a linear growth in your capabilities while actually decreasing the resource base that you need and I

5:47 think that's the kind of the fundamental framework that that operators need to start to adopt start to adop this

5:54 thinking from the hyperscalers and apply it to our networks so let's look at how do you begin to apply that what are the

6:00 principles and the the key findings from our research and we came back with these

6:08 five kind of key findings and they are not you know they are overlapping right so these are not entirely complimentary

6:14 they they do work across each other the first one um first kind of design and

6:20 operating um principle that that we found is think flexibility so from the

6:26 outset because we don't know how demand will evolve and we don't know how those patterns so think flexibly the second

6:32 one is you know focusing on how we minimize new resources and those new

6:38 resources are energy but there're also new equipment resources and there're also other fabric buildings um and and

6:46 um energy and and those resources I think another key finding that came across quite heavily from our

6:52 conversations was simplification the need to simplifying things both from the

6:57 customer perspective but also within the operator and and that giving us more

7:03 resilience better reach and and also improving security kind of interesting is how many

7:09 of the people who spoke to talked about you know building the network with AI in

7:15 Automation in mind so AI enabl Automation in mind so starting to think now not necessarily about implementing

7:22 AI overnight but thinking how do I make this network automatable how do I make

7:28 it how do I prepare for an AI enabled world and then finally there were some really interesting

7:34 findings about unlocking value in existing in new fiber and and how you

7:40 know that fiber estate and in some markets there is considerable fiber and

7:46 in other markets it's more of a scarcity but that also significantly influencing

7:52 how we think about um Network design so these five principles I'm just going to drill down on each of these quickly in

7:57 turn so on the first one here Planning and Building flexibility on the left here we have the um as it were the old

8:06 world right so um on the left it's how a lot of networks were designed um and and

8:13 built and to some extent that is still true today we still you know we've still designed networks to deliver on demand

8:20 content consumption streaming um applications U from the cloud so there

8:26 is still a fundamental flow of traffic that we are supporting um and and that's

8:33 how we've been thinking about our Network designs to date with a more centralized architecture and with more

8:39 predictable traffic patterns we think although we don't know and and it's you

8:45 know who are potentially more distributed because we actually don't know um but we think that there's going

8:51 to be a more distributed traffic flows we've seen this already and L from light stor talked to this um that that the

8:58 there's a lot of traffic between data centers has emerged um which is quite different sort of flows of traffic um

9:06 and also you know much more localized much more distributed and we need to

9:11 match those traffic flows with more distributed architectures um to deal with that in a more resilient kind of

9:17 way so that's the first key principle that came out of of the work the next

9:22 one just spend a little bit of time explaining this is about the focus on

9:28 emissions so for those of you who are not familiar when it comes to reporting carbon emissions we talk about three

9:35 buckets of emissions we talk about scope one which is the direct emissions from

9:41 combustion of fuel um so it would be the you know burning of fuel for for

9:47 vehicles the use of generators that's scope one so all emissions relating to those activities then there is scope two

9:54 which in the case of telecoms is primarily around use of electricity so it's one of the emissions that are

10:03 indirectly um produced um through the consumption of electric power um and

10:09 that's a fairly significant um domain it's it's manageable um and but it's a

10:16 very significant part of the overall um emissions and then there's scope three

10:21 and scope three is for all operators it's the biggest component of emissions

10:26 and that includes all of the other um supply chain um embedded carbon

10:34 within the networks and so that's everything else excluding electricity um

10:40 and excluding um combusted fuel but everything else that that you have to pull together to deliver your service as

10:47 an operator so those are the key ones we're looking at increasingly operators are focusing on that challenging

10:55 significant scope 3 but there is still plenty of need to maintain the focus on

11:01 scope one and particularly on scope two now you know we're somewhat simplifying

11:07 things but how do you think about minimizing those resources and those resources can be energy resources which

11:13 are a significant cost but they can also be that supply side in other words the component resources how do I minimize

11:19 the carbon embedded carbon in my supply chain and the principles we we were that were discussed with us one of those is a

11:26 more modular design and that the notion here is if you can build a more modular design then the overall um Hardware uh

11:35 life cycle is extended because you don't have to upgrade everything you can upgrade what you need to when you need

11:41 to and and that you know that modular design allows for um a a much more

11:48 incremental renewal rate but also allows for a much longer life cycle of your equipment the second one um is the

11:56 second element of that is modular design also it helps um to to reduce the blast

12:02 zone of Vault so it has a maintenance um benefit too the second key principle is

12:09 about more converged architectures and that's you know that's the ability to

12:15 avoid duplication avoid duplication of resources avoid duplication of teams and

12:20 of equipment and have um a more converged architecture that can be convergence on a service level but it

12:27 can also be a convergence on network layers and we'll come back to this point in due course I think you know a key

12:34 thing that we've heard a lot from speaking to operators is this need to the physical constraints um and

12:40 certainly minimize the physical footprint of plant and Equipment um and do not extend that physical footprint

12:47 because that's concrete and that's buildings and that has a a big KnockOn effect on your scope two emissions um

12:56 sorry scope three emissions so some sign ific physical constraints that operators

13:01 need to maintain um and ideally minimize and even reduce their footprint and then

13:06 finally you know AI also has a role in optimization and it has that role on the

13:12 active infrastructure um and we we all know about those kind of sleep modes within our network but it also have a

13:18 significant role to play in optimizing the passive infrastructure our heating our cooling um and our other estate so

13:25 um those are key principles for for minimizing the use of resources within

13:31 um and meeting our journey and the next finding from the research and just move on to that quickly was around um

13:40 simplification for the customer and that's about um having much simpler Network designs borrowing from the

13:46 lessons of the hyperscalers where yeah hyperscalers have done a lot to drive down and simplify their architectures

13:53 within their data centers within their digital infrastructure can we learn from that how can we adopt um some of those

14:01 Concepts and principles um the second key benefit of simplification is it allows us to scale more effectively so

14:08 it gives us the flexibility to scale it gives us um what we need in terms of

14:14 responding um quickly and I think the third benefit that was highlighted in our conversations is if you simplify it

14:21 reduces the complexity and and also reduces um points of failure um and

14:26 therefore it can improve your up time and overall performance so those are the the key elements of that and then

14:32 finally the the well two more elements I just want to cover off quickly I'm conscious of time um on the next slide

14:40 we have on the next slide yep we have the

14:45 um use of AI and for autonomous networking and that's you know that's on automating the network it's on creating

14:53 you new generative IIA user faces but it's also about automating the passive infrastructure and G Management Systems

15:00 as you can see from this quote here and it's also dealing with

15:07 increasingly AI generated customer inquiry so I think you know AI is a

15:12 great tool for us um as an industry but we need to be cognizant that increasingly you know we're going to be

15:19 dealing with AI as a customer um and is thinking through about if AI a customers

15:25 giving AI agency how do we better support um our customers who

15:30 increasingly will be adopting AI for making networking decisions that we're providing them and and how do we also

15:36 ensure that our own AI agents satisfy things like SLA so there's a whole new

15:43 world of of operating models that we need to step up to finally the last of

15:49 the the findings from our research was about existing fiber and we had some really interesting conversations here

15:55 with operators because some operators are in geog Ries where there is lots of fiber and when there's a lot of fiber

16:02 that does change the way you approach your network design because it suddenly means that you can think about doing

16:10 things differently um and you can you can start um revisiting your network

16:15 architectures you can start going down this simplification route um and that's going to be quite different by geography

16:22 so but what we did find is that you know there's obviously more and more fiber available and as there's more fiber

16:28 available within the network so to you know you can you can think differently

16:34 about the network and it's less of a scarce resource and you can start you know adopting different principles when

16:40 it comes to design where there's still scarcity then you know obviously that

16:45 that that changes it but there were some interesting findings there about patterns of of Supply on fiber and how

16:52 that influences the thinking on the network right I'm going to then move on because i' taken up enough time on on my

16:58 presentation so summary these are the five key findings what I like what we're going to hear now from Amit is how how

17:04 is um you Juniper um thinking about it well providing capabilities to operators

17:12 um along the lines of our findings so over to you

17:17 em hi everyone uh hope you can all hear me thank you uh i ph I think that that

17:23 was awesome to talk about all the different things you guys have done the research uh with the customers

17:29 um the fundamental issue really is that the the as the demands continue to grow

17:36 and AI is going to make those demands even more exponential from a growth perspective we need to S get to the

17:42 sustainable uh uh also achieve the sustainability goals you know for the networks uh I want to talk a few things

17:49 that um pH mentioned just repeat this for for a minute because there's a point I want to make as we go to the AI part

17:55 of this uh we we know the traffic is growing right we can debate how much the

18:01 traffic is growing but uh during the co it was growing much faster then it

18:06 slowed down a little bit and now it's going back to the normal rates of growth right it's it's it's

18:12 exponential um video has been the driver for many many years uh even after all

18:18 the humans start watching the video that they can watch and saturate that we we will still have a new demand coming from

18:25 machines who will be the watching videos especially in the AI era to do some real-time decisions so this trend of

18:32 video growth will continue for a very very long time the businesses uh what

18:37 they are doing is they are putting all the applications in the cloud so literally everything that a business is

18:44 doing needs to access a cloud which means all that data growth is coming

18:49 from the business side too on the right side the two big things is about data center right uh because

18:56 everything is hosted in the cloud so so the traffic in the data centers and between the data center countes to grow

19:02 exponentially and then AI is going to add an exponential piece of this because now we start to see very very large

19:10 models and these large models shares share a lot of data between the different gpus that sit within the data

19:16 center and across the data centers uh so with that I want to talk about AI on the next slide before we go to the some of

19:22 the product pieces stuff let's go to next slide so

19:29 uh what does AI mean for networking there are two concepts I want to talk about one is networking for AI and the

19:37 second is AI for networking but before I do that I want to compare this AI ERA

19:43 with what we have seen in the last 20 years which was the internet era what did internet era do it basically

19:50 democratized information right so everybody could access information in an inexpensive way then we had 5G which

19:58 allowed accessing that information from any place anywhere any time uh then we B

20:03 got Cloud which made uh the information storing and accessibility in do in an

20:10 inexpensive way right uh by bringing economies a scale so this cloud cloud

20:15 plus 5G uh plus uh DC era uh brought a

20:21 lot of new innovation right we got e-commerce we could watch video anywhere we had Co but we could still work from

20:26 anywhere now we to the AI era what AI era I believe is going to do is it's

20:32 going to democratize intelligence which means people are going to be able to access intelligence in an in inexpensive

20:38 way uh it will drive new business models right we talk about things like you know

20:44 uh uh automated driving or self self full self-driving kind of Concepts we

20:50 talk about extending GL life human life sense personal assistance and even extending the human brain if you are

20:56 born to be futurist with the man machine interface but as we come back to see what this means from a networking

21:02 standpoint the first phase of this is what you see on the right where the training is happening right the training

21:07 of the large models is happening in the Big Data Center and it's driving large amount of traffic within the data center

21:14 across the data centers right um this is so massive that it eclipses everything on the left of it today but that's not

21:21 where the money is that's all cost training of models is all cost the money is going to be in the inference and

21:27 let's stay the pr slide Bill the money on training in is in the infrance and the insurance is basically where which

21:34 is going to happen at the edge of the network and it's going to happen in the access access of the network and even

21:39 may happen at the customer Edge this is the money right as we look at this uh

21:45 from a community perspective I think service providers do have a big opportunity here while looking at

21:50 inference that how do they monetize because all these traffic for infrance is going to require SLA and this traffic

21:57 is going to sit are this these capabilities are going to sit at the edge of the network uh next slide so as

22:03 we look at this some of these Trends from Ai and from the video next slide please uh we will see different choke

22:10 points we we are seeing right now choke points at the right end of the network which from the DC perspective we are

22:16 also seeing choke points at the left of this network which is in the access space because the new fiber is still

22:23 being deployed I live in the middle of Silicon Valley I still don't have fiber to my home and I can believe this is

22:29 true for many people on this in this audience um so we have this choke points when these choke points get filled up we

22:35 will have inference coming to the edge in the core and then those choke points will come up so we will see choke points

22:41 at different places Network as different kind pieces of time and the fundament thing we want to talk about is how do we

22:46 how to address these right in a sustainable way the number one thing is performance which we call as flexibility

22:53 to scale uh a silicon and and and the new speeds and feeds play a big part here

22:59 the number two here is sustainability the sustainability is not just about

23:05 doing lowest cost per bit uh it's really about doing the right tool for the right

23:10 job because in for example in the core lowest cost and lowest power B is the

23:16 most important thing but at the edge of the network lowest cost for service is the most important thing so you need to

23:22 measure the TCO in the right way in the different parts of the network and use the right tool for the right job and the

23:28 p is pie of operations and this is where going to start looking at AI in terms of how what we do with AI for networking to

23:35 simplify some of the processes of automating the networks uh next

23:43 slide so when we look at the implementation of this we start with the very fundamental concept we start with

23:49 the fundamental concept of what is the cicon we need to drive to solve the

23:55 different problems of the network at Juniper We Believe we need to use the right tool for the

24:01 job you cannot have a Toyota engine in a Ferrari and a Ferrari engine and a Toyota it doesn't work and some other

24:08 the guys believe one silicon but in Jer we believe you need to do use the right silicon uh so we are investing in three

24:14 classes of silicon we are investing in MX routers which uses a trio silicon and

24:20 this is the most flexible silicon in the world because even the Silicon which we shipped 10 15 years ago we can drive new

24:27 capabilities in San today right because it is fully programmable uh on top of

24:32 that it has the scale which drives the lowest cost per customer and per service

24:39 so this is optimized for the edge and it has to scale in terms of cues number of filters uh number of services you can

24:45 you know can terminate this is the best in the class on the planet then we have PTX product family which you uses

24:52 Express silicon and Juniper invented this class of silicon and some of the other guys followed that

24:59 and we invented this class of silicon because we wanted to bring lowest power and lowest cost per bit in the hypers

25:06 scale bandwidth uh you know locations which is which we call as core networks bearing networks or data center networks

25:12 right so this is uh the Silicon that's focused on that and then we're not always Bound by our silicon we do use

25:19 merch and silicon and we believe that merch and silicon makes sense in some parts of the network especially in the

25:24 access and aggregation space because here we looking at you know smaller form factors um and silicon doesn't play that

25:30 much differentiation there what's most important there is the software and the capabilities that we drive with junas

25:37 and D using different Merchant silicon to for the smaller form factor Platforms in the access and aggregation space so

25:43 all in all our play our strategy really here is to offer these multiple options to the customers so they can pick the

25:50 right option for the right place in the network next

25:55 slide and this is basically showing uh all the different uh portfolio this this

26:01 is doesn't show every product on the slide because that would probably buy too much so so we're going to say we

26:06 have the PTX router um now shipping with 800 gig um and these are optimized for

26:13 the corent puring applications and aggregation in large aggregation applications for service providers MX

26:20 with the trio silicon both in the fixed form factor and the uh model form factors uh optimized for the edge for

26:28 the service terminations and HX for Access and aggregation with the smaller form

26:33 factors on the automation side uh we have Paragon solution which drives all

26:39 the big pillars of Automation and also the AI op uh and I'll talk about that in a few slides uh on the Optics side we

26:46 have invested uh in our own Optics when it comes to uh the Cent Optics uh

26:52 currently we are shipping uh 400 gig Optics and we will soon be having 800 gig Optics also so uh for our new a gig

27:00 Platforms in the PTX family next

27:06 slide so uh I want to talk about this a little bit which is what we call as the coherent Optical routing or we what we

27:13 used to call as IP over dwdm you know in the past when we did IP over dwdm we

27:19 always lost density on the routers but that is no longer true every port that

27:24 we have on PTX MX ACX this could be 100 Gig port this could be a 400 gig port or

27:30 this could be 800 gig Port is capable of doing a a coherent Optical interface

27:36 just with an optic plugged into that Port so as you look at the two two

27:41 scenarios in the past you had to deploy a router you had to deploy a transponder on monder and bunch of optical equipment

27:48 uh to basically build a network that's no longer needed and and this is being

27:53 driven both in the service provider space and in the data center space

27:58 with integration of coron Optics into the routers because first it drives much

28:04 better economics you know less power less pace and less C footprint some the numbers are on the slide and these are

28:10 very very compelling numbers second it drives a lot of Simplicity because you're eliminating so much stuff from

28:17 the network and less points of failure much more stability and visibility into the network I think this this is a

28:23 no-brainer and as I look at some of the numbers uh how this technology is taking off the technology is already taken off

28:29 in the hypers scaler world but over time when we when we look at the numbers from the analyst then the numbers show that

28:35 the service provider volume will become bigger than the hyperscaler numbers in terms of these Optics because um it will

28:44 be in every part of that I said could be it will be 100 Gig 400 gig 800 gig and in the future for 1.6t and 3.2t

28:52 also uh next slide uh just a quick quick when you

28:58 talk about sustainability it has to come with the numbers I will not read all the numbers but I give you some examples um

29:06 why we drive uh silicon every 3 four years because every as we drive to the nextt Silicon it does give us a better

29:13 power footprint right so for example today we are shipping our new PTX routers uh which is uh based on the

29:20 express fire silicon we are seeing 75% power efficiency as compared to the

29:26 previous generation and that that's true for MX every generation and that's also true for ACX which is driving better

29:33 space and power as compared to the competition even by using that same silicon because of the software on that

29:39 um so I think this is the key and as we bring the optical pieces cerent pieces that basically icing on the cake because

29:45 that drives even additional power savings to build the sustainable networks uh and and the both silicon as

29:52 well as Optical piece is key along with the packaging of this from a product perspective

29:58 next slide so the the product is one part of

30:04 that the the second part of that is how do we run the the networks in the most

30:09 optimized way right uh so in our Paragon solution we have the four big pillars

30:16 planning orchestration assurance and optimization but what we're looking at this is when we looking at this we

30:22 looking at this from a perspective is how do we basically Drive different use cases with that because the use case

30:28 Drive the outcomes so I'll give you a couple of examples from this slide here first example I'll say is when I onboard

30:35 something can I unboard something really really fast so can I can make money and revenue faster right so that that's

30:41 driven through Paragon through zero touch provisioning onboard stuffs much faster but really over time you you

30:49 onboard it you provision it most of the time the automation tool is going to be optimizing the network this optimization

30:56 comes in two ways uh to scenarios okay there is a closed loop piece so you take up take care of your uh scenarios of

31:03 failures uh which can be automated uh the the close loop automation of those

31:09 or you can look at this optimizing how the networks are working uh and this is where we bringing Innovation from a

31:15 green networking standpoint because we've added many knobs to our platforms

31:20 uh which can be manipulated by the automation system which is Paragon to run the network in the most Green Way

31:27 what does do that mean if something is not being used we will turn it off um if

31:32 something is uh you know there is a big difference between day and night so we can drive lots of savings between how

31:38 the network is run in the day and how the network is run in the Night by turning on and off things to run in the

31:44 most power optimized fashion next

31:52 slide uh I think now you might have heard about Juniper Miss so Juniper

31:57 acquire this company which has been driving U AI

32:02 into Wi-Fi space into wir space and into the now we're bringing that even to the

32:08 routing space uh especially this is true for many Enterprise customers who if you

32:13 have deployed uh a Wi-Fi and and a campus and Branch network uh with the

32:20 same zero permiss we can provide routing Assurance in the van Edge for the T

32:26 routers so uh integrating this AI Ops because our routing platforms provide

32:31 all the Telemetry um and this Telemetry can be used by the Z permit to provide

32:37 the writing Assurance just as the Wi-Fi or the wi Assurance has been provided so in the Enterprise space that gives you

32:43 the single tool so now we have two capabilities we have Paragon which does all soups to nuts from an automation

32:48 standpoint and using jiper mes into the Enterprise space especially providing the AI Ops along with as they have been

32:56 using it for the V and um Wireless space next

33:02 slide so this is my last slide before we go to the the panel uh if I can

33:08 summarize this there are fundamental pillars of innovation right how we can build sustainable networks it is a

33:15 silicon and jber is driving the best silicon both in cost per bit and cost per service

33:21 perspective uh we drawing a lot of efficiencies not in not only building the right model andix form factor

33:27 platforms are also giving the knobs in those platforms to run those platforms in the most power efficient way uh so we

33:34 got full portfolio with PTX MX and ACX the every port our routers is capable of

33:40 doing dwdm 100 Gig 400 gig or 800 gig and automation is basically that's taking all those capabilities and and

33:48 running the networks uh in in a green way because by turning off and on knobs with uh at the right time based on the

33:55 usage in the network itself so that's my last slide and and then Phil

34:01 we can move on to the panel discussion thank you very much Amit um

34:09 and I'd like to invite the other panelists to switch their cameras and their mics back on and I would also like

34:14 to remind everyone who's um who's joined as an attendee to submit your questions

34:20 we've already had one or two questions com in which is great um and I would invite you to submit more questions we

34:27 will try and get around to the questions um in a moment if we run out of time and

34:33 we don't have the questions we will endeavor to respond to your questions that didn't get answered today and and

34:40 therefore send out the responses after the event but let's just kick off um and

34:45 I'd like to just start with our panel with you what do you see as the three

34:51 big fundamental drivers of change um that that are driving um Network

34:58 transport architectures and thinking and and how will these impact operators so

35:04 actually I'm going to start maybe um with terj what do you see as the the

35:09 three big drivers of fundamental change in networking yeah thanks Phil so so I

35:15 think actually basically it's almost covered in introduction and also by amid but I can I can maybe uh shortly repeat

35:22 as as I see it and then of course diving into what does it really mean for us as we see it so so so the when it comes to

35:29 the three three big drivers I think the number one which was clearly addressed a couple of times is uh the growing

35:35 traffic and we can argue about the number of course the number of annual traffic growth depends on what countries

35:40 you are in and also what kind of Technologies you you're holding out so last uh last years at least in our

35:46 footprint as Tor in in the across the different countries we see roughly something between 20 to 40% annual

35:53 traffic growth uh not to rest on those because we see of course when we're launching new technologies like you know

36:01 launching 5G and those kind of things there will be a jump in the traffic so that's the kind of one of the key driver

36:06 is of course the growing traffic on on in general uh and then is of course this

36:11 traffic needs to be handled with lowering the cost so that's also I think important I think amid mentioned that a

36:18 couple of times so so and also remember that the cost is composed of capix and Opex so so the operational side of this

36:24 is very important in that sense uh the second part of is that we I think also actually amid covered that it's

36:30 customers want more they want more than than all call it only connectivity so so

36:36 so they want actually a solution and experience they can trust so that means that they typically want you know more

36:42 bit manage services they want a bit more on security they may want more on insight into selfservice they want

36:48 support of AI and those kind of things so so so that also has a consequence on

36:54 on the capabilities we have to implement in the network and also the support system and thirdly is that also I think Amit

37:02 walk through that this networks need to be automated they need to be secure they need to be green and they need to be

37:08 resilient uh and then of course that has a direct consequence on the how we are implementing this uh with the modularity

37:15 with robustness uh capable of supporting this for example introducing all relevant

37:22 security mechanisms and that part of it so that's kind of the the key drivers I think so if I may summarize is uh it's

37:29 simply you know high capacity simplified architecture I think you call it converged architecture even F earlier

37:36 modularity uh we also need to support you know personalization uh we I think two two

37:43 more things I think would add to what the what you mentioned earlier I think first of all resilience I think we we

37:49 need to be capable of of uh of supporting any kind of scenarios in

37:55 particular when we are providing you know critical infrastructure uh and we see that customers are moving on also

38:01 both Miss Mission and business critical applications onto that uh and the last part of it which is probably a new one

38:08 uh is U we need Network architecture to be able to engage with customers in a

38:14 different way uh then we call that experimenting innovating co-creating and those kind of things which is also

38:20 additional kind of requirements when we see coming from the digitalization across the industry call IT industry for

38:27 zero or whatever but it's a different engagement then with customers uh and and then we need support from from the

38:32 network in order to work in that way that was very short I think on the on the drivers and and some of the aspects

38:38 we see as consequence of those well thank you for that um am and and I'd like to if

38:45 maybe if you feel you can L um can you condense um your response would you add

38:52 anything to what just said in terms of the fundamental drivers I mean you have a different perspective you're a new

38:58 operator with with a very um interesting sort of business profile what would you

39:03 say from your perspective the big drivers of fundamental change in network transport are yeah I think the number

39:11 one for for me would be open Line systems uh you know open Line systems

39:16 are becoming more and more if we have to get the coherent Optics to work you know what amth was was positioning and you

39:24 know more and more coherent Optics uh is going to be available now so uh you know

39:30 they they would basically require an open Line system uh we are seeing a lot of disaggregation of subsy cables so the

39:38 subsy cables of the past and the subsy cables of future completely different uh we have L more fiber Pairs and uh the

39:46 the PFE and the slts are you know disaggregated you can you can have the slts in data centers and build uh you

39:53 know every service provider can have their own pair of fi C they can run their own protocol systems uh and

40:01 services on it which is very very different uh you know GTM model than

40:06 what was prevalent and the and the third one I think is you know a lot more

40:12 people are trying to simplify the transport Network and do away with uh uh

40:17 you know uh with with with dwdm any protocols to run on that you know so so

40:23 have no control plane you know just completely use it as a simple line system and build all the intelligence in

40:30 the routing plane so those are the three things that I would say okay great well I'm going to move that thank you both

40:37 that's really interesting and it does sort of tie into what we heard earlier I'm going to move to Ahmed on the next

40:43 question which is um you know what are the primary sustainability objectives

40:50 for operators in other words for yourself arment and for photoone and how are you addressing this in in terms of

40:56 evolving the network and the architectures the primary sustainability

41:02 objective for The Operators like WaterOne uh is reducing carbon emissions

41:09 and converging uh Energy Systems and minimizing electronic waste uh and this

41:16 is could be bu by uh energy efficient modules reducing power consumption to

41:21 minimize environmental impact and operation cost uh resourcing optimiz

41:28 ization just like efficiently utilization network resources such like

41:34 a space and power electromechanical infrastructure totally circularly economy adopting practical that promote

41:42 uh the ReUse and recycling of materials by dismantling the obsolete uh networks

41:49 and uh ICT loads to address this objective we must uh deploy energy

41:55 efficient equipment with high efficiency uh and low uh power

42:01 consumption implementing virualization and automation reducing the physical just like I said dismantling the old and

42:08 obsolete infrastructure and Management Systems exploring a renewable energy

42:14 sources instead of uh a utility sources great well I think we're going

42:20 to come back because there's a question that's coming um from the audience um

42:25 which I think is is interesting because it relates back to what you're saying ared and I mean the

42:32 question um that the the the questionnaire is asking is you know do

42:39 we have any good examples of success stories with how the value chain can

42:45 work better and collaborate to deliver um these sustainable outcomes so can I

42:51 ask you briefly to respond to that ared and then I'll I'll hand it over to terj

42:57 yeah yeah it's a good example for uh sustainability is dynamic seral management uh this is a AI module

43:04 implemented on the data centers to enhance the cooling system uh to be more

43:09 sustainable and uh reliable for the ICT loads uh by reducing uh reduce and

43:18 increase the the fan speed relative to uh the heat dissipation required for the

43:25 ICT okay great and you you got examples of where you've worked across the supply chain for that kind of outcome and I

43:32 just touched back on you tday do you have any case studies where where you've

43:37 worked with the supply chain and collaboration any success stories you could point

43:43 to yeah just to give you a couple of numbers maybe Phil if I may as a starting point to give you the context

43:48 so so if you look at Tor you know the climate you know Scopes and how how is

43:56 that split into different different contributions is exactly what you said I think in your presentation

44:01 that we have roughly 80% I think the last time I checked coming from scope three so that means that you know

44:07 looking at and collaborating with suppliers is is one of the key elements when it comes to the to that

44:13 calculations uh and then if you look into the the energy consumptions roughly 80% or or even 90% s of that from from

44:21 our own Tor consumption comes from the network uh and then of course networks is composed number of domains space

44:28 stations transport mobile core data centers and so forth and then then we have the further split into that but I

44:33 think it's important not to get lost in this you know split into the domains you you need to look at the total Network

44:40 solution uh so we actually we're doing that I think it was two years back uh

44:45 and we we summarized you know the achievements that we have been through at that time and then we uh managed to

44:51 toh keep the energy consumption roughly flat despite a 200% traffic growth and

44:58 that's breaking away exactly from or you have to break away the energy consumption from the from the traffic

45:03 growth that think curved and what what how we did we do do that I think is kind of the question next of course it's uh

45:09 so we we Sunset you know all Technologies uh calling 2G sorry 3G for

45:15 example copper Network and those kind of things uh we replaced some of the older Solutions with newer Solutions mean

45:21 being coming more energy efficient meaning uh you know more efficient chips

45:27 for example utilizing you know sleep modes and and lower clock rates and all those kind of things so that's the kind

45:32 of the tricks uh uh in quotes we did in order to uh to achieve that flattish

45:37 kind of development and all these kind of things uh required close collaboration with suppliers uh because

45:44 they need to do their part and we can do our part of course yeah okay well some interesting statistics there and that

45:50 that separation of resources from outcomes um really good example I'd like

45:56 to come back Amit to you um you you mentioned several times in the

46:01 presentation um that Juniper's recently announced it's adding 800g coherent Optics and pluga bles into your

46:09 portfolio we did actually discuss this with some of the operators and the research and and to be honest they

46:16 didn't see the need for that kind of capacity in their Network today um so

46:21 what's the thinking here okay thanks for asking that question Phil so I I I give you some

46:28 examples first of all um we we are shipping 800 gig today right on the

46:34 routing platforms uh we are already working with 50 customers across the globe to uh to

46:40 get to 800 gig deployment uh and let me just address

46:46 the 800 gig specifically that go to 400 gig and 100 Gig the 800 gig is not just

46:51 800 G right I think that's the misn number sometimes what is what does 800

46:56 gig to the table 800 G brings a much better IP over dwdm solution as compared

47:01 to the 400 Gig Generation because the 800 gig Optics D Optics can go even uh

47:08 not just Metro and Regional distances but also the Long Long Haul distances

47:14 right so that's one thing second 800 gig is also 2x 400 G it's a much it's a it's

47:21 a uh it's a it has discrete fiber ports it's not a breakout it's a much better to 2x 4 gig you get a much power better

47:28 power footprint and then if you take it go down it's a much better 100 Gig solution also because you get a lower

47:34 power footprint with this NE class of silicon so 800 gig is basically giving you many different things better 100 Gig

47:42 better 400 gig and better IPO ewm solution it's not just a 800 G which is

47:47 what some people might be thinking about and then I want to take out the dwdm question also because when we talk about

47:53 the dwdm question dwdm question is not 800 gig wave length I think now we are

47:58 at a point where we can do 100 Gig 400 gig 800 G wave length on every port on

48:04 the router because there are Optics which are um optimized for 100 Gig 400

48:09 gig and 800 gig which really really makes things very very flexible okay great so that responds to

48:16 some of what we talked about the flexibility the scalability but also realizing that the same

48:24 technology is very flexible it can can work across multiple speeds and and L I

48:30 mean I'm interested in bringing you in here because you're dealing with a lot of speeds you're dealing with these huge connections between data centers

48:36 explosive growth and and I know you've looked U you're obviously looking very closely at this um uh and where's your

48:45 thinking at regarding um the use of speeds and the Technologies and coherent

48:52 Optics so clearly know uh the the routers where you would need 800 gig PS

48:59 uh you know if you look at the traditional uh data center networks it's uh you know typically they would at Max

49:05 would ask you for a 400 gig Port uh 800 gig is a unique uh kind of an ask and

49:11 you know if if you have to really look at you know what's really going to come at you are these uh Nvidia servers so

49:19 one Nvidia server let's say h h100 it can take up to 3.6 terabyte of so so one

49:27 server requires that kind of capacity that itself will take 45 800 gig FES so

49:32 the what's going to come at us very soon and and that requires a completely

49:37 different technical skill set in a networking organization deal with those

49:43 kind of speeds coming out of a single server and you know they have clusters of these servers and the Clusters have

49:49 to talk to each other so the leaf spine models are going to fail they're going to be completely new uh new designs

49:55 going to come up grounds up so I believe that this technology has a great uh

50:00 future you know when it comes to uh making AI uh you know uh democratization

50:08 of AI you know that's that's going to happen and you know that's going to change the uh the Dynamics uh but you

50:14 know if you go and talk to any of the traditional Telos and you know we are now four years old you can call us traditional now uh we we have a lot of

50:22 customers who do od4 Services ODU one odu2 services and those Services uh have

50:29 to be now circuit emulated in into you into the coherent Optics and

50:36 you know the question is will those uh will will will the o build those kind of

50:43 capabilities into into the routers or not so I think that is kind of a question I would have for a

50:50 myth yeah that's a good question so let me start with saying I I do um have

50:56 great respect to the fact that service providers do have existing contracts of providing certain set of services it's

51:02 not easy for them to turn the contracts overnight like hyperscalers cod right so so there is a operational uh challenge

51:10 in service providers face to move to a new technology uh but I do believe uh

51:16 that like you said earlier open Line systems so many service providers are already adopting openline system that is

51:21 the step one to adopting the current technology in the routers uh

51:27 there there will I believe there will be a hybrid approach for for a period of time where you do your lay one Services

51:33 um because you have the business in the layer one side but everything else from a routing perspective you can put domate

51:39 that coron Optics onto routers and drive those lay Lins in a cheaper way over time I do believe as the speeds and

51:45 feeds and the routers go grow in the service provider space uh the same 100 Gig service that you're providing today

51:51 using web Lin uh you can provide that using ethernet private lines and give the same experience to end customers and

51:58 same cost and power so I to totally understand why most ofice prod is today

52:03 but we we are seeing the transition and I think once the transition happens this

52:08 is why I said we seeing analy reports that over time the service provider current volume will become bigger than

52:14 the hyperscalers and U and I'm banking on that too myself I think that will

52:19 happen I me with the combination of that prediction and what L's just been talking about the kind of number numbers

52:26 were talking about with AI the growth the phenomenal growth and the demand for

52:32 that and the need to fundamentally rethink the network architectures to serve that um you know that seems to be

52:41 a very significant impact coming down towards the industry are we ready for that is the service provider Community

52:47 because up till now networking you the AI Community has focused on silicon and

52:54 they've been obsessed with you know Nvidia or whomever are they now going to have to start you know figuring out

53:01 networking constraints that they haven't to date you had to worry about so

53:07 much yeah I think that's a question for the panel I think this trend is coming and L said it's starting from the data

53:14 center and and anybody in the business of providing connectivity between the data centers is seeing it today and then

53:20 as the inference moves into the to the edge of the network those demands will

53:25 move to the edge of the network and that's where the money is you know a lot of people are going to make money in infrance and it's a great opportunity

53:32 for Telos of the world to figure out how to monetize uh this

53:37 transition great I'm conscious we haven't got much time I'd like to put another question here this is a question

53:44 from Peter and he's asking you know great Innovations and cudos for all the

53:50 work done he's saying whereas you seem to be addressing both the environmental and the governance issues in the Telo

53:56 industry I would like to hear more about the societal factors mostly how employees and staff within the company

54:02 how do you engage them so I'm to come back to Ahmed and I'm going to ask you know your experience how do you engage

54:09 the organization to also sort of Step Up and support the initiatives around

54:15 delivering this future sustainable network with this colossal growth how do you how do you engage the rest of the

54:21 organization

54:28 you are you able to share the experience in vone for that yeah yeah yeah I'd like to say uh how to engage uh

54:36 my company in this uh in that uh uh

54:42 Innovations ideas uh by make a full calculations and assumptions for uh

54:49 using uh automation modules by uh four modules capacity management Performance

54:55 Management environment Dynamic Thal management Inventory management we make a full

55:01 uh Gap analysis and what if analysis this that's why AI modules produce uh

55:08 the Gap analysis between the uh current consumption and the forecast consumption

55:14 and uh what is in the key performance indicators uh uh insights with the

55:21 energy gaps so uh I prepare uh uh a lot

55:27 of uh innovation ideas like uh UPS's in

55:33 enhancement efficiencies D station enhancement efficiencies cooling the

55:39 cooling is the most partical uh of the energy uh consumptions in the data

55:45 centers uh so I prepar all the Innovation and make

55:51 the the assumptions for the some kind of savings in in cab in opics as

55:58 operational course as capital courses so I uh prepare my uh project

56:06 or uh The Innovation idea and uh bring it to the management system through uh

56:12 with uh uh with the the correct predictive uh savings and they say

56:22 yes great so it's it's using the exist management and governance and supporting

56:28 around that um and it's it's also I think sharing that with a wide audience I'm sort of conscious we are we are

56:34 short on time so I'm going to just pick um the last question here maybe um which

56:41 is um there's a question here which is interesting from Ranny the ecosystem is

56:48 moving rapidly and generating scope three emissions that are rendering Net Zero targets unobtainable multiple

56:54 hyperscalers have had to renounce on their previous carbon reduction targets how do we as an industry ensure that we

57:02 don't fall into the same trap of the hyperscalers and um and and that that

57:07 that we are able particularly for our Hardware related emissions how do we enable that we are able to stay on track

57:14 so very quickly um how do we do that you've got you got a a minute each I'll

57:19 start I'll start with you TJ how do we how do we not fall into that disgrace

57:25 that the hypers scalers fall into I think if you look into the climate of course it's energy source and energy

57:31 consumptions and you have to look at both of the sides there so so so you have also have to engage you know with the ones producing energy uh mean

57:39 meaning renewable energy and that's what you're doing for example through Power purchase agreements and those kind of

57:44 things uh but then also we are working very intensively on on Energy Efficiency

57:50 across all Network domains and that also of course requires to go into the uh into the supply chain in order to make

57:56 that happen so so uh back to the one the just the question you had before this

58:01 has to be shared across you know the whole taor immediately across all the units to make the best make use of the

58:07 best insights the best practice how to do that so so that's the way we are doing that of course uh we have set

58:14 Ambitions and and are aiming for delivering on those this is a great discussion I'd like to invite the others

58:21 but I'm I'm afraid we have run out of time but we got going we got some great questions coming so so those of you who

58:27 submitted your questions we will try and get back to you and we will share the results of the comments with everybody

58:35 um so um we will get back to you sorry apologies we not being able to answer all the questions great submissions from

58:40 the audience it leads me to thank the panel um for your contribution and to

58:46 thank um Juniper for sponsoring this event and for their presentation and wish everyone a good day thanks everyone

58:54 thank you thank You