Light Reading Webinar: The Open, Automated and Programmable Transport Network

0:06 hello everyone and thank you for attending today's webinar the open automated and programmable transport

0:12 network sponsored by sienna infinera juniper and fujitsu

0:18 before we begin i wanted to cover a few housekeeping items underneath the side view is the q a if

0:24 you have any questions during the webcast you can type your question into the q a box and submit them to our

0:29 speakers all questions will be saved so if you don't get 20 we may follow up by

0:34 email at the bottom of your audience console are multiple application widgets you can use if you have any technical

0:40 difficulties please click on the yellow help widget here you can find answers to common questions

0:47 a copy of today's slide deck is available for download in the green resources widget

0:52 towards the end of today's presentation we'll ask you for your feedback a survey is already open on your screen and will

0:58 only take one minute to complete your feedback is extremely helpful an on-demand version of the webcast will

1:04 be available about one day after the event and can be accessed using the same audience link that was sent to you

1:10 earlier today i would now like to turn the event over to heavy readings senior principal analyst

1:15 sterling perrin thanks becky and hello uh good uh good

1:22 morning good afternoon everyone my name is sterling paran i am with heavy reading and i will be

1:27 the moderator for today's webinar on open automated and programmable transport network

1:33 uh this is a part of a special larger project um as noted sponsored by sienna

1:39 fujitsu infinera and juniper and we're really excited today to

1:45 share uh actually a great deal of survey results from the project that are just

1:50 now being made available um this is the first um

1:56 launch of any of the results uh here's a look at the agenda for the next hour

2:01 and then i'll also in the next couple slides i'll give a little bit more background on the project as well as the

2:06 demographics for the survey and introduce our speakers but once i do those introductions and

2:12 introduce the project we'll go into the the meat of the survey results and have a quite a bit of discussion among our

2:18 panelists on the topics you see here timelines and requirements for automation in the

2:24 transport network cloud automation specifically uh quite a bit of of data on optical

2:30 transport in automation and then moving into optical line systems including open

2:35 line system and we should have a good 10 minutes for q a at the end so please hopefully there's a lot of questions and

2:41 hopefully you ask them and we'll get to as many as we can the project itself we do a number of

2:48 these market legion market leader projects and they have a similar format of course

2:53 different topics this one is on automation for transport networks and just to give you a sense of scope

2:59 that includes transport pretty broadly defined packet layer transport ip and ethernet optical

3:05 layer as well as a lot of focus on multi-vendor multi-domain and then on

3:10 multi-layer moving across packet and optical layers the basis of the project is a global

3:16 survey we completed that survey in may so it is very fresh results as i noted

3:22 uh it's a global survey 78 respondents in the end i'll give a little brief view of the demographics of the

3:29 the survey in a moment of course the readout is today on this webinar so thanks for joining uh

3:35 there'll also be a white paper uh coming in july i'm just putting the finishing touches on that actually hoping to glean

3:40 a couple additional insights from this webinar and put them into the paper and then there'll be a set of analyst blogs

3:46 written by me over the coming a month or two demographics um

3:54 larger operators represented uh typically for our like reading surveys as well as a geographic split of we we

4:01 shoot for about half north america half all other regions you can see the breakout there for this 100 of them

4:07 involved in network planning and purchasing or poor purchasing also all of the companies we included

4:13 are either using or planning transport automation so we really did want people kind of in the know of

4:18 of automation for transport and already had some views uh to to land on the topic so it does skew a bit more towards

4:26 um uh you know people involved in in transport automation which is great

4:31 here's our panelists i'll introduce them briefly and then we'll we'll hear from them quite a bit as we go through the next hour kent jordan representing siena

4:39 advisor of portfolio marketing uh also joined by theresa monteiro director of

4:44 marketing with infinera joe masserino principal solutions architect in the packet optical

4:50 networking um group within fujitsu and then kenica atree senior director

4:56 product management strategy and business development uh representing juniper today

5:01 so welcome all and as i said we'll we'll hear from everyone in a in a bit actually quite a bit

5:08 let me kick off the content with timelines and requirements um

5:14 looking at uh first starting with the importance of of

5:20 automation for the transport networks one of the basic questions we asked right up front how important is

5:25 automation for your next generation transport strategy it was really quite an amazing result from my view 93

5:32 of the respondents view transport automation is at least important for their next generation transport strategy

5:37 important being uh defined as automation will be used extensively but there are other factors more important

5:43 and then also combined with really uh an astounding number on critical uh 39

5:49 automation will be primary pillar of the of the network going forward we ask this type of question a lot in different

5:55 surveys and to get at 39 and critical is quite um

6:01 very unusual so this is showing that we're really hitting on an important topic here no doubt

6:08 uh time um different timelines uh so on the left is

6:13 the global results on the right is the group uh you know if you look that group that said automation was critical we've

6:19 actually carved them out in a lot of the data as um as um uh priority adopters or early

6:27 adopters and just seeing how they respond uh so this one on the global results of about a third of them saying

6:33 or exactly a third currently deployed or planned to deploy in 2022 so you know

6:38 fairly aggressive but when we carved out that priority group the 39 that said it's critical

6:45 74 of them have either already deployed automation or planned you by the end of the year so again that gives us in

6:51 looking at those results through the survey it really gives us a little bit of a sneak peek i think of where the market is going ahead

6:58 and we'll do that a bit here and also in the white paper itself um

7:04 so let me uh just kind of break into the discussion here um use cases for automation this is one of

7:10 the early questions we asked as well which of the following transport network use cases are you planning to automate

7:17 within the next two two years and you can see the list i've organized here in descending order traffic engineering

7:24 emerging uh as a top priority but also you could see at over 40 response provisioning network

7:31 inventory and configuration also scoring quite highly uh let me bring in

7:36 kanika from juniper to kick off the discussion here i guess first thoughts on how the automation use cases came in

7:43 and then secondly in terms of your own customer discussions how does the results here fit with what you hear from

7:49 customers thanks for inviting me to comment um

7:54 well first of all i'm not surprised at all given the kind of demographics that you shared are mostly people who are

8:02 already using some kind of automation or they know it's very critical and they're investing in it that's why i see some of

8:08 the more advanced use cases like traffic engineering top off their list uh to your question on you know how

8:15 we're seeing the customers um response to some of the use cases i mean i as

8:20 part of my uh job in uh you know owning the whole portfolio for automation for juniper

8:26 i talk to lots of customers and you know they don't do automation for automation's sake they they're doing it

8:33 to drive business outcomes and when i ask them what are the business outcomes that you

8:38 want to achieve basically say two things one category says well i want automation

8:46 to help me simplify those daily manual repetitive

8:52 tasks and make them so scalable repeatable high quality that i can close my eyes and rinse repeat them right so

8:58 that's one category of use cases stuff like you know software upgrades stuff

9:04 like activations testing provisioning and you'll be surprised even there you

9:10 know there's a huge role uh even for those who've adopted automation there's a huge scope to actually improve the

9:16 total time it takes right your software upgrade today takes days weeks sometimes

9:23 and then the second category of outcomes that my customers are telling me is really those complex things like even if

9:30 you throw more humans at the problem you won't be able to solve it right these are things like assuring your network

9:36 and service performance your kpis you know optimizing capacity autonomously

9:41 reducing the risk and that is sort of the second category where you see use cases like traffic

9:47 engineering where you think see things like predictive uh analytics right so i

9:52 think this reflects the responses you see here it reflects a very healthy mix

9:58 of both categories of use cases we hear about and it reflects also the stage at which

10:05 these operators are you know they always start with the easy manual stuff and then they go to the more complex things

10:13 which is also a function of how much traffic and you know how big their network is and how complex it is

10:20 yeah good good points um on the um the bit of a skew of of the

10:26 audience that responded because we did kick out companies that did not have the

10:32 five-year window to to implement automation we did not include them so this is more which is typical of our

10:38 surveys it's a bit more of an advanced view of the overall group because it's kind of self-select people that are

10:44 interested in investigating the topic already we'll want to respond um and then second of that though although i

10:50 don't show the breakout here but then when you take that critical or early adopters group then you really get the

10:55 leading edge i don't think um the leading edge group was

11:02 tremendously different from here because i didn't provide a breakout but but there were some some differences um but

11:08 this is kind of a mix of everybody and this is what what you get um joe any um

11:13 thoughts or comments uh just to bring in joe from fujitsu any thoughts or comments you wanted to make

11:19 in terms of the the grouping and breadth of use cases here uh sure uh thanks sterling and in fact

11:27 uh one of the things that i noted was technologies required in order to

11:33 implement you know these levels of automation you know key to this use case

11:39 automation is optimal orchestration automation enabled by open apis

11:46 for optimal use case operation essentially you need a combination of

11:52 optical orchestration sdn domain control analytics and mlai

11:59 and and through these uh technologies all working together you achieve that that level of

12:05 automation that that zero touch operations that uh kenika was also

12:13 reflecting in results that she was seeing all right excellent we've got of course

12:19 a lot more detail we'll go into so just one final note on this for the audience so this was

12:25 as i said holistically a survey of transport automation this was a general question automation broadly defined what

12:31 are the use cases we will dig into different areas of automation and actually i think

12:37 revisit the the use cases a bit in in different areas as we go into cloud automation or cloud cloud-based

12:43 automation in a moment and then we talk quite a bit about the optical layer uh and there would be differences but this

12:48 was the the broad brush automation and transport what are you planning and it is a nearer term window of the two years

12:55 let me uh share one more slide here on the more general broad um

13:03 uh trends and and then from there we'll dig into the cloud automation specifically uh

13:08 but on this one of course it's always important as an analyst to ask about what are the barriers to it to adopting

13:14 uh technologies and uh though the one that came up top here what are the primary barriers for

13:21 automating uh your transport networks uh generic automation frameworks slow to tailor to our needs

13:28 um ranked as the chief barrier selected by forty percent and then you can see there's really another three that um

13:35 that bubble to the top as well which are fairly closely

13:40 um aligned in percentage of response lack of in-house software expertise

13:46 execution complexity uh cost too high lack of budget though those tend to be the the top um

13:52 i don't have the breakout here but uh i did uh break out uh in in parsing the

13:58 results the priority adopters and they do have a slightly different um

14:03 skew of versus the the non-priority or kind of the not to say laggards because

14:08 nobody's quite allowed at this point but the the others group um the priority adopters really keying in on the generic

14:15 frameworks as their top challenge uh interestingly and not surprising when

14:21 we broke out by the others group the non-priority group

14:27 lack of in-house software expertise was the top one as well as lack of budget so you get a sense that this priority group

14:33 has moved past they've hired the right people they've got the budget and now what are they running into and it's this

14:39 frameworks uh challenge i don't um i'm not going to go into discussion on

14:45 this one yet i'm just putting it out here but i think as we get through other areas of discussion i think we can kind of revisit the slide in some of these

14:52 points uh as as we go through uh and i think one of those areas where the challenges may come back

14:58 uh is as beginning a discussion of the cloud automation which um i'll move into now

15:06 and um yeah so again this is getting kind of drilling down into into specific areas

15:12 so in this case we asked um does your organization plan to use

15:18 transport network automation applications from the from the cloud and here what what i did is is give the full

15:25 global results on the left and then did one of those early adopters or priority adopters versus other breakout to to

15:31 show some differences uh on the left you know public and

15:36 private hybrid uh 52 percent saying saying they would um use cloud in that fashion 31 percent

15:43 fairly high saying private cloud uh from the global resource results a bit of hesitancy to go to

15:50 straight public cloud only interestingly when we looked at the

15:55 early adopters which the orange bar on the right uh tremendous interest in public and

16:01 private hybrid 76 and much less private cloud only among them as you

16:09 compare that to to the others group so you can really see i think kind of an evolution of of

16:15 thought and maybe comfort uh kanika i know you and juniper were instrumental

16:20 in some of these cloud questions um thoughts here

16:26 i'm actually very uh positively encouraged to see these results it's synonymous with what we are hearing

16:33 and it kind of dies back nicely into your slide around barriers to adoption

16:38 if you if you think about some of those barriers right generic automation framework

16:44 not enough software development expertise too complex they are all interrelated and they're interrelated to

16:51 that idea of many of the operators trying to kind of build their own organic diy

16:58 monolithic systems that somewhat are analogous to boiling the ocean

17:04 and what's happened since is you know we actually saw another survey which

17:09 said 80 percent of these diy projects they kind of failed you know failed in

17:15 terms of delivering the first business outcomes and their management challenging the roi hey show me the

17:21 returns right so that mentality now we are seeing the management uh you know asking those

17:28 questions and saying hey instead of boiling the ocean go fast but start

17:34 small start with one problem you want to automate show me the results

17:39 go to the next and when you want to do that cloud is the best delivery model to get

17:45 there because you can get started with no overhead you know you literally just hook up your

17:51 device and you select one use case you don't have to go all in

17:56 and you know you can start running that automate just that for example device onboarding or ztp or just inventory

18:03 right pick one from that list just automate that but show the outcome in

18:09 three months versus three years so that's what we are hearing and that's one of the rationale why we are actually

18:16 seeing some of the forward-looking you know operators who've kind of been there done

18:21 that burned their hands and now they want to you know reap the benefits of the cloud

18:27 yeah interesting um that connection between the challenge of the generic framework

18:34 which which again really really came through uh in this survey and as you noted from your own um separate research

18:40 in conversations to to um how the cloud can help let me um

18:45 advance this one here and kanika i'll keep you on uh to to weigh in here so this was one where

18:52 we took those use cases um for automation and said which which ones are of primary interest or cons to be

19:00 considered first for public cloud-based uh automation and transport and you can see the audience can can see

19:06 the mix here um um i guess my my one observation i'll let you comment is my middle bullet the

19:12 cloud usage it did show a fair but not exact alignment with the general use case priorities um

19:18 any any thoughts um here on on on how these cloud use cases came up and and

19:24 how they compare and and just any um any more thoughts and then i did want to bring in another topic with you as well

19:32 yeah absolutely i mean if you think about these use cases there are mix of management plane

19:39 control plane and of course data plane so we're not obviously touching the data plane at all but

19:45 if it's really just about observing a network that's a little bit more management

19:50 or touching the network changing a piece of config here and there there is a much greater comfort level of the operators

19:59 to do those use cases from the cloud but when it comes to control layer use cases

20:04 right in the future things like network slicing um or traffic engineering you know it's

20:10 not everybody i mean that's where maybe our conversations is a little bit different from this survey and i'm

20:16 actually happier to see these results here but we do see some hesitation there

20:21 where oh okay in control plane right like is it gonna add a latency you know where are you going to

20:28 render it from so there's a lot of things that need to unblock and eventually i believe all these use cases

20:35 will be cloud delivered that is no doubt the future but i do believe there's a crawl walk run to to that stage and we

20:44 will start initially with the easier sort of management plane use cases

20:49 yeah i'm just just thinking on that traffic engineering because we've been discussing that that that a bit um

20:55 offline um you know i'm thinking it through so in my discussions uh segment routing is

21:02 definitely very tied to to traffic engineering and that you know we call that sdn we certainly call that

21:09 automation we are i am seeing a pretty strong pickup and interest in moving with

21:15 segment routing and it is coming through in other surveys i imagine when we talk about set traffic

21:20 engineering here perhaps that is part of what a lot of operators have for their um segment routing uh would you agree

21:28 absolutely absolutely absolutely like the sdn controller particularly shift to

21:33 sr is critical to this uh whole piece yes yeah and there's a number of operators

21:38 i've been talking i've just been surprised at how quickly that is moving forward from

21:44 again this is a bit more skewed towards more advanced companies generally so it makes sense that in context um

21:51 so juniper has and you've been talking about an automation as a service concept so i wanted to let you weigh in on that

21:58 to close out this section and i believe you do have a contributed slide there so i'll let you introduce the concept

22:04 define it and really just you know why is this important what's the value

22:10 yeah thank you for the opportunity uh so you know i personally and at juniper you

22:15 know we all believe that the future of automation is delivered from the cloud

22:21 uh you know the guiding principles are first speed you know automation is all

22:28 about moving at the pace of business not at the pace of internal systems and today that is one of the big barriers

22:35 right and cloud is all about moving fast it's you know quick time to market

22:40 second guiding principle is you know automation should let you focus on productivity not on production right

22:46 like so there should not be any overheads to kind of maintain manage the whole infrastructure hardware software

22:52 lifecycle you know the 25 different components from open source from vendors from your own you're basically

23:00 heavily you know becoming a software development house yourself rather than focusing on solving network

23:08 problems you are you know there's an overhead and cloud solves that i mean

23:13 that's just not sustainable right and thirdly it's the easy button right like there's a whole lot of simplicity with

23:19 cloud that you can really get to with you know with complex uh in-house

23:25 systems it's really hard to kind of make it simple and consumable so we believe the key success factors for automation

23:32 number one in the future is going to be delivered from the cloud that's why we at juniper are doubling down on it

23:38 in addition to that this whole idea of use case based right go small but go

23:43 fast and then you know move from there right don't boil the ocean

23:48 and um you know the second part is ai enabled ai is going to be an

23:54 intrinsic part of the whole cloud infrastructure the data layers the training layers are going to

24:00 be built in into that cloud infrastructure and the value of ai is multiplied you know 10 folds

24:08 when the data is coming from all the different sources which means that a problem discovered in one network in

24:15 north america would be solved even before somebody in apac sees it you know

24:20 and again all of this is anonymized so the power of ai is basically multiplied

24:26 by cloud and we believe it makes sense to do ai ops only when delivered from the cloud so that's the second key

24:33 success factor the third thing you know we believe is as part of any kind of automation solution assurance active

24:40 assurance is really critical it's almost like saying well i spotted an issue i

24:45 figured out why it happened i fixed it but hey did you go back and actually

24:51 check it worked and that's the idea of active assurance so you need to actually go back and

24:56 check and verify everything and this idea of verification has to be you know built in into automation um

25:04 at juniper we're doing that both at a hardware level as well as a software level right

25:10 and you know our active assurance solution has a component built into embedded indoor routers where

25:17 you can basically your router becomes an assurance agent right and then the last idea that is very

25:23 important for success is security and trust and that also is both you know at the

25:28 hardware uh level so at juniper we believe you have to have you know things like tpm

25:34 2.0 uh those kind of unique identifiers built in into hardware to make sure

25:40 absolutely unbreakable you know hardware out there and at the same time

25:45 top it up with as a service security trust all those things to make sure that

25:50 this entire cloud operation is absolutely uh you know safe and secure so

25:56 we are uh thrilled to drive this future uh and uh you know we'll be delighted to

26:01 talk more and you'll hear from us uh more on this very soon

26:06 all right excellent thank you um and yeah we've done some other survey data that that does uh have

26:13 some fairly strong supporting data behind uh some some of what you just brought out there so looking forward to

26:19 how that progresses let's move on um into

26:24 the the next section so we're moving into the optical transport portion we had a number of questions here

26:30 uh in the survey and so let me uh just start introducing them here and

26:35 i think we'll have some interesting discussion around these as well uh so you know really starting at the top uh

26:40 typical question we ask in our surveys what's driving automation and specifically here for

26:46 open optical networks uh what are the most important drivers for managing and automating your open optical network and

26:53 again you can see the percentages and how they came in we put them in descending order capacity visibility and

26:59 resiliency really bubbling to the top maybe bring in sienna to kick off this

27:05 discussion kent um i believe you were quite interested in this question in the survey sienna was

27:11 um thoughts on how this came back any observations yeah so these uh results really like you

27:17 said they're showing that service providers are looking for automation to really help with some of the higher

27:23 level functions in the optical network things like optimizing their capacity or improving the visibility of the network

27:31 getting better network resiliency and i can provide a few examples of how some of those things are being delivered

27:37 today so there are apps that exist to help optimize capacity across c l band

27:44 based networks that can provide time series analysis of performance measurements

27:50 from modems to be able to help service providers optimize their

27:55 capacity and if you look at optimizing things like spectral allocation within the fiber

28:01 that opens up opportunities for new applications like spectral defrag

28:06 where you can then further optimize the fiber utilization and sienna's been developing apps and features in a number

28:13 of the areas on this slide for quite a few years and network operators are

28:18 now using these tools and applications in their optical networks today with

28:24 i would say kind of the lead applications for automation uh in this area are really around

28:30 providing uh real-time visibility into how the network is operating

28:36 which is something that you know they didn't have the ability to see before and you know over time we expect these

28:43 types of applications to be used uh in new ways and in more sophisticated ways

28:50 all right excellent um i know we had a fair amount of panelists that were interested in this topic joe

28:57 uh any additional thoughts or comments on on this one sure

29:03 optimizing capacity and fiber utilization uh through automation provides just more

29:09 accurate results than the manual tuning methods the manual performance optimization uh

29:15 requires highly skilled people expensive test equipment and also manual uh

29:21 measurements uh and tuning takes just too much time uh to implement uh and

29:28 then there's also the the human error factor in collecting uh those results

29:34 um so having that that complete automation uh capability uh within your

29:41 system operations really in improves on the efficiency on the delivery of your

29:48 overall services and also regarding end-to-end network visibility and oe m automation

29:55 will simplify that uh the system operations

30:01 for life cycle management and future service editions all right excellent uh last word if

30:07 there is um teresa from infinera to bring you in um but if

30:13 we've covered it all that's fine too any final thoughts on this teresa yes i have i have two observations and

30:20 thank you for having me here um sterling um first of all i do understand and and

30:27 we hear it from customers the interest in optimizing capacity and making the best out of the resources but if i read

30:33 the title of this question um drivers for automating the open

30:38 optical network and if i focus on that open then i think the point of end-to-end network visibility is a real

30:45 critical one because when you're moving from a closed system where end-to-end network visibility is a given and you

30:52 move to open you you worry am i going to lose that end-to-end network visibility

30:58 so um i think that's uh an interesting point to keep in mind the other observation i would like to make is well

31:04 you you have there a bullet that says automated system turn up is secondary

31:09 but i've actually i was playing with some of the survey results and i filtered the results for this question

31:16 for the participants who are focusing on the optical technology

31:21 and what we see is that for those the system turn up the simplified wavelength

31:27 turn up the automated calculation of the coherent transponder parameters become much more relevant and because at the

31:34 end and they would uh go to levels as high as 45 percent or around so because

31:40 they're the ones who are undertaking their tasks therefore they understand the benefits that um automating um the

31:48 service provisioning um bring my my company good good points on both

31:56 um you've got automation of the transport network generally and then open and then keying in on open house

32:01 other kind of two to or opens a subset of overall and also a good point i had not done that filter

32:07 that was a good one uh there's many ways we can filter these results uh that i kind of keyed in on the early adopters

32:13 but that's that was a great nuance to add let me go to uh so this next

32:19 question on um optical layer functions what functions do you need from the optical layer

32:26 um and so this was kind of more general um uh

32:31 we got some strong results at the top two there ability to move capacity flexibly and then um

32:38 ability to easily connect new locations and minimize truck rolls etcetera terry assistance we've got you on um i'll let

32:44 you kick off off this one and and you know feel free to agree or disagree or add to

32:50 what i kind of bulleted out there thank you sterling so i was really happy

32:56 to see that um these results confirm our understanding that the optical layer is no longer this

33:03 static uh over-provisioned um connectivity or pipe uh set of pipes

33:11 basically where other layers run into this is not enough anymore operators need to scale fast they need to use

33:18 their resources more efficiently and they need to enable new business models bandwidth on demand network as a

33:25 service for that they need on one hand the programmable in infrastructure that

33:30 allows them to move the capacity flexibly um but but they also need the software that

33:37 supports that and and i'm going to move actually to the next slide um because

33:44 oops there we go sorry oh thank you thank you because that's actually what we call

33:51 instant bandwidth at infinera the support for pre-deploying hardware

33:56 hardware that has a very large capacity pool that can be increa incrementally

34:01 activated you activate it with a click where you need

34:06 when you need and um you pay as you grow so i think this is the type of automation solutions that um

34:15 operators are looking for when they talk ability of moving capacity around

34:20 ability of turning up new services with as little truck roles as possible

34:27 all right excellent uh let me go back on this um and then kent um

34:32 we do need to move along because there's a fair fair amount in this section but ken if you have some thoughts or observations here i'll let you actually

34:39 close this one out and then we'll go to the next question okay yeah i'll try to shorten up my response but i i agree

34:45 it's great to see that this question shows that network operators really understand how

34:51 the optical layer can help them develop a much more automated network it's not

34:57 just us that believe it but i think that you know this result shows that they believe they're not looking for a dumb

35:02 pipe optical layer that's only providing connectivity they want some value-add from the optical layer whether it's

35:09 additional intelligence resiliency ease of use or flexibility and a lot of that

35:14 can be delivered like was already said through a combination of hardware like rotom

35:21 layer zero control plane software architecting hardware so that you can minimize truck

35:28 rolls and new applications like one that we have which is called liquid

35:34 restoration that can be used to provide more resiliency without additional investment in the network and it does

35:40 that by identifying additional restoration paths for each home path in the network

35:46 including paths where you know wavelengths could be restored maybe at a lower rate to partially restore services

35:52 and help maximize you know the resiliency of that existing network all right excellent let me go to the

35:59 next one we're gonna we have a lot of good data but in the end there might be a bit too much for

36:04 one hour um managing multi-vendor networks another interesting question kent since you're

36:09 on i'll let you continue here um which of the following are part of your goals for end uh goals for managing your

36:15 multi-vendor transport networks and again you can see the percentages descending thoughts and comments on this

36:21 slide yeah so multi-vendor as you can see from the results it's a big topic and you

36:27 know it's exciting to see that there's a lot of interest in many of the the responses

36:33 sienna's working with operators in this area who want to be able to make sure that their network architecture is open

36:39 and doesn't impose restrictions on the technology and whose technology that they can deploy so typically we first

36:46 will look at things like visualization and control being able to visualize a third-party

36:52 optics across a cnn align system using an open interface like an open config interface

36:58 uh to discover topology between that transponder and the sienna line system

37:03 and then being able to control that transponder provisioning and pushing parameters into

37:08 it some of that may actually require some development resource adapter development

37:14 for example being able to use those open interfaces to set things like frequency launch

37:19 power and so on and then after we've kind of got that base visualization and control we look

37:25 at things like streaming telemetry to pull in alarm data pms from the device

37:30 to track metrics over time maybe look at specific look for specific alarms to be

37:36 more reactive and so i think we see that open apis visualization control and

37:42 analytics troubleshooting these are all important aspects of open multi-vendor networking

37:49 all right and um joe i think you wanted to to comment on this one if you can do a quick comment

37:56 uh because we've got two more in this section i want to hit hit and then cover the last section

38:01 sure in fact just drilling down into some of the technologies involved in

38:08 attaining uh you know these types of results advanced transponder

38:14 dsp technology using mlai in conjunction with the orchestrator to

38:20 simplify execution of these functions this essentially enables a real-time

38:26 capture of your optical layer metrics which then can be used uh to automate

38:32 the network uh capture uh the characteristics of the fiber uh to

38:38 optimize uh your service performance end-to-end all right excellent let me get we've got

38:45 uh yeah so then we wanted to address uh planning tools and and again keyed into

38:50 open here what type of tool you use for planning services over in open optical networking uh open optical network uh

38:57 teresa i know you contributed several of the questions or see on the infinera side

39:03 around the uh open optical thoughts or comments on this one

39:09 yeah i'll try to be i'll try to be quick and as you mentioned their operators strongly favor planning tools supplied

39:16 by an equipment vendor but on the other hand they they want multi-vendor planning tools which is all fine if you

39:23 think we're talking about open optical um planning and i think um i think

39:29 actually the the concern here and and this comes from many discussions with uh

39:34 with customers is that we as vendors are not yet doing our best to um share

39:42 details on our devices and and share them in in a uniform and consistent

39:48 manner so that so that these tools can have the

39:53 information required in particular if you want to plan networks that require high optical

40:00 performance but i think um operators are doing their uh their job doing a great

40:06 job there and pushing us vendors to go more into this

40:12 cooperative or competitive cooperative mode of uh information where

40:19 we walk the walk of open basically yeah interesting interesting point and

40:25 it does show fairly clearly that one 41 versus 28 percent is the second that uh

40:31 it is fairly strong uh finding uh let me go to the last one

40:37 go ahead sorry i just wanted the vendor basically owns the the planning solution

40:44 then and that's a better ownership than that of open source where the ownership

40:49 is uh undefined right the open source was was quite secondary there but but still some

40:56 interest uh for sure last on this topic and then we'll we'll move into the line system and then i

41:02 think we've got a little bit of debate within that section so i'm quite curious how that plays out machine learning um

41:08 so looking specifically at machine learning's impact which functions will benefit the most from machine learning

41:14 in an open optical network and three tied and it's a bit unusual but you know these are you can't get

41:20 more equal than that you could see the top three network design predictive health predictive traffic growth as the

41:26 top opportunities uh theresa to close out this section um

41:33 uh is that consistent with your discussions and and really what infinera is planning

41:38 itself it is very much consistent i would say there's lots of

41:45 complex problems in open optical where machine learning can help

41:50 offering faster solutions solutions that are more accurate maybe than the deterministic algorithms that exist

41:56 today but the killer use cases are still a bit open in the air there's lots of

42:03 discussions and proofs of concepts and we have actually a lot of white papers published

42:09 on machine learning for quality of transmission that's related to the to the

42:14 first item that you have there also machine learning for preventive maintenance

42:20 predicting the network health um also there so i think those topics are still

42:26 maturing let's say until we find the ones where there's really an advantage

42:31 in using these technologies predictive traffic growth on the other hand um i don't think it

42:38 has been talked so much in terms of machine learning obviously there are some studies

42:43 but um and i i think that's maybe because um the old um

42:50 traffic growth models that operators have been using for metro or for long haul may be changing with distributed

42:58 computing and coherent coherent going down to the edge maybe now there's a larger uncertainty

43:06 and more concern on how are the traffic patterns going to to change

43:11 now um the self-tunable transponders i'm i'm pretty sure and actually i've looked

43:17 into it if you also again select the the participants the

43:22 respondents who are um focused on optical technology that also becomes

43:28 another contender for uh machine learning as as significant as the others that that would be my remark

43:35 interesting interesting i'll certainly parse it that way when i um complete the report part good observations let's move

43:42 to the last section uh we are running a bit behind but um

43:48 joe on the online system i know fujitsu was interested in several of these topics i'll let you kick off this one

43:54 the most important drivers for deploying a new line system this was my view curious what what what

44:00 your thoughts but you know certainly savings and capacity are the typical drivers that we see in optical

44:06 networks to move forward open architectures came in third but not a weak third it's it's it's definitely

44:12 third but it wasn't the primary driver thoughts on on how these played out joe

44:19 excellent points uh that were mentioned here and i i essentially see two technology areas that address uh these

44:28 uh important drivers for deploying a new line system uh essentially new amplifier

44:34 technology and the use of both c l bands in the new amplifier uh

44:41 techniques uh that are offered uh minimizing relative intensity noise or

44:47 rin uh reducing the cost per bit per kilometer

44:53 along with offering higher gsnr this means a higher order modulation

45:00 essentially more bits per hertz can be used within the spectral with enabling

45:07 and overall fiber capacity increase or better utilization of the fiber

45:13 use use of the l-band although the the l-band has uh traditionally

45:19 not been as cost-effective or as widely deployed as the c-ban on uh the rotom or

45:26 line system its strong transmission characteristics have made it an ideal uh complement to

45:34 c-band for expanding system capacity and instead of using add-on modification

45:41 to the existing optical line system a new continuous c

45:47 plus l system using a single wss

45:53 in the rotom will simplify operations uh eliminate any potential blocking

45:59 essentially also providing a major cost savings uh by uh basically having the

46:06 amount of equipment needed uh to implement the optical line system

46:12 yeah good good point on cnl and incapacity we're certainly seeing that as well let me uh

46:18 go to to this one and then we've got two more after this to close out uh operational challenges joe i guess we'll

46:25 let you uh hang in to weigh in on this one the most difficult operational challenges when planning deploying

46:31 anything greater than a two degree uh optical line system uh really yeah big big one uh showing here lack of

46:38 modern planning tools is a major challenge this quite clear finding um i'm assuming it's consistent with what

46:45 what you're seeing but um what do you think absolutely and in fact to overcome

46:52 these challenges we begin with open apis along with the combination as

46:57 i mentioned earlier in the webinar optical orchestration sdn domain control and intelligent optics

47:05 offering analytics and mlai this combination will offer

47:11 real-time wavelength optimization uh simplifying planning installation and

47:18 ongoing operations uh this combination of technology will also

47:23 reduce the dependence on highly skilled technical staff through uh efficiencies

47:29 uh that can be implemented as well all right excellent and then let's get into where

47:34 two of the the interesting i mean there are many interesting ones um i particularly found this one

47:40 interesting and i think you guys did as well uh moving into open line system

47:45 architectures um which are operators planning to deploy 47 saying open line system with a single

47:52 vendor with multi-vendor uh coherent optical network in the interworking this is

47:57 really a mix and match of transponders being preferred by this group um and then second but fairly distant open line

48:04 system from a single vendor with a bookended approach um i'm a bit well somewhat surprised by this but it is

48:10 fairly clear kent uh from uh sienna i'll let you

48:16 give the uh kind of the leading view on this one um thoughts and comments and um

48:22 is it was this a surprise for you yeah i think these results are pretty interesting i mean they show that

48:28 there's a lot of excitement and willingness to deploy multi-vendor

48:34 coherent optics and so i'm a little surprised at the result without knowing more behind the data because you know

48:41 it's uh typically much easier to deploy that middle option and we've seen several service providers that

48:48 have made announcements that they are deploying that uh partially disaggregated model and we

48:54 have many customers that are deploying that that one that's in the middle of the chart so the partially disaggregated

49:00 bookended transponder model of course you've got things like oif 400 zr

49:06 with interoperability that's available so that could be driving some of the result but ultimately

49:12 moving to that top option may mean some extra integration work that will need to be done and at

49:19 the end of the day you know from a vendor perspective we support all three models and we have customers that are

49:25 interested in all three we have coherent optics that do offer that interoperability

49:31 um whether you want to do that or the bookended transponder but i think it's going to be interesting

49:36 to see how many operators actually do adopt and deploy that first option over

49:43 you know the second option yeah good good point teresa i don't know if you did your your optical specialist

49:50 analysis on this one but um you want to weigh in on on how these came in

49:56 i do so um optical experts um support the fully disaggregated open

50:03 line system with full interworking only only six percent of them support that

50:09 approach and i think i mean that's clear it's them who have to make it all work

50:14 and come together nicely so but thank you for sharing

50:20 so the overall company may it may have interest but the optics group is saying this is going to be a

50:26 tough a tough yeah interesting joe any any last word on this this slide then we

50:31 have one more uh that's another interesting topic to hit sure sure and and we can see from the

50:36 survey results that network providers are still split on what shade of open to use and you know

50:43 single vendor ols multi-vendor coherent optics it it it's optimized for metro uh

50:50 you know metro works best in this environment you you still require interrupt testing uh but then

50:57 there's also the service impact of uh troubleshooting challenges uh that one

51:02 can expect on on that second point uh a single vendor open line systems with

51:08 bookended transponders i see this as optimized more for long distance networks uh having that book ended

51:15 transponder where uh the effect long distance performance

51:20 uh is achieved through you know that those bookended optics

51:25 all right excellent let's uh close out with this one um which could be it's and has been its own

51:32 topic of a full webinar but on this one what are the challenges managing coherent plugable optics um which is

51:39 really referring to the ipo over dwdm architectures and the biggest one at least for the full survey group was

51:45 maintaining the existing operational practices of these um separate groups um

51:51 joe i'll let you close out on this topic

51:57 thank you uh while economically appealing uh to first class uh you know

52:02 the uh it essentially opens the door to adversely impacting existing operational

52:09 practices uh affecting mttr performance qos and potentially

52:17 service availability as well as the cultural and organizational challenges uh that tend

52:25 to weaken business strategy i i see you know we see a way around this uh

52:30 as this challenge it's to essentially coordinate the the optical and ip

52:36 controllers uh in that the optical controller manages pluggables like a

52:41 transponder end to end uh the optical controller uh provides open apis uh like

52:48 tapi for example uh to the layer 3 controller and then the layer 3

52:53 controller maintains ip routing with knowledge of the end to end optical

52:58 connection all right um you know what maybe ken i'm just thinking that uh on infinite side

53:05 on on sienna's side that you've certainly done a fairly broad um

53:12 foray into the plugable optics given and also given your comments on the last question any insights um to to add on on this

53:19 particular one and then we'll go to q a yeah so i think uh you know i don't want to downplay these challenges because i

53:25 think some of them really are challenges and you know many network operators

53:30 service providers are definitely seeing the potential benefits of converging their ip and optical layers instead of

53:37 operating in traditional silos however you know many of the platforms so the

53:43 the routers and switches uh that the coherent plugs are being put into don't allow or enable open northbound

53:51 interfaces to allow third-party management of that plug so if you don't access that

53:57 plug you can't manage it and you need basic information to be able to manage the network like it's done today

54:03 without taking a step backwards and as an industry you know we've made a lot of advancements with respect to open

54:10 opening up the line system you've seen a lot of that in the presentation today gaining visibility to what's happening

54:16 in the network and if you put a plug in a router you don't want to lose that capability management needs to be

54:22 on par with what you're doing today you know where you're where you're doing things separately so

54:29 i think the second piece is in addition to that base management network operators also

54:34 want to take advantage of any of the dsp technology advancements that are going into those plugs and again not having

54:42 access to the data that's coming out of the you know performance data that's coming out of those modems in those

54:48 plugs is a barrier to getting to that detailed optical performance analysis which is

54:54 one of the the challenges here so getting direct access for the controller

55:00 uh to the coherent plug through an open northbound interface enables both the base management of that

55:07 plug and also lets the operators get you know a maximum roi on their optical

55:12 investments so i think you know these challenges there is some work still to be done here

55:18 right right these things don't happen um you know a lot of times that the timeline is the wishlist is a year out

55:24 but they tend to take a lot longer for these big mega trends let me um we have a few minutes left for q a um we

55:31 did have a lot of content there's actually more in the survey and this is just a fraction of it so there's a ton of data

55:37 in here but let's get it to a couple of the questions um uh maybe uh kenika let me um there were

55:45 actually a few questions that did come in around the automation side early on let me bring you back in

55:52 um question are there specific wan dominance when domain slash roll is there a

55:59 specific when domain where automation is more critical than

56:04 other parts of the network so i guess where where is um specifically in the network where is

56:10 automation going to play most most heavily in your view

56:15 yeah i think that's a great question uh i mean specifically if you look at the traffic patterns metro is the home uh

56:22 it's kind of the new edge right uh that's where all the service distribution is going to happen off from

56:28 that's where much of the traffic uh patterns are changing you know easter

56:34 east west the complexity the scale is growing so you know we think automation build out

56:41 uh with cloud metro goes hand in hand really well the other

56:46 new domains where i feel like you know where the next new technology introduction is happening that's all

56:53 going to be automation led meaning automation not as an afterthought but

56:58 really right from the start examples being i think integrated optics is one of them i think we talked a lot about it

57:05 right things like scale out um you know fabric scale scale out

57:10 fabrics in in the van uh site um png cups and and you know some domains

57:16 so some of the new domains uh but right now i mean i definitely see metro as the biggest opportunity to go

57:23 in with automation first all right action let me go to um this was an interesting one um and

57:31 actually i've done done a bit of work on this um question came in uh will digital twins

57:39 help the automation process um joe i'll let you you um give at least

57:46 the initial response on this but i do think it's an interesting question george thank you

57:52 darling so yes absolutely delivering digital twins of the network helps

57:58 simplify a full life cycle of system oe m and future service editions

58:06 optimizing capacity improving uh network availability and essentially minimizing

58:12 your your capex all right excellent um yeah we i did a

58:18 white paper around uh just just that very thing specific to uh research man

58:24 management which is one of the early identified here early use cases of interest uh

58:29 question on um and we got about two to three minutes

58:35 left so uh for the audience you know you you can certainly ask your questions if we don't get them uh our our panelists

58:42 are going to be happy i'm sure to follow up individually but we do have some great questions in

58:47 um how do you prioritize combine the development of the programmability

58:52 of the network versus development the new features in your own network management system interesting and i'm

58:58 sure relevant for many ken i'll let you let you take this one yeah i think this one's yeah yeah i

59:04 think this one's pretty important because automation is increasingly a key business challenge that service

59:10 providers are focused on and from a development perspective you know operational features and

59:16 functionality is a key aspect that we as a vendor are focusing on developing but

59:21 we've got to you know closely align the programmability flexibility

59:26 capability of the hardware and software on the network elements with you know the capabilities that are

59:33 being developed uh on the network management system and

59:38 even things like being able to do uh testing more quickly by like let's say

59:44 doing your integration testing in the cloud can help with some of this speeding up you know adoption of new

59:51 um features and functionality for automation so you can actually do

59:56 now testing on virtualized network elements that are running the same software that you would have in the real

1:00:02 network so if you've got you know testing being done in different countries or time zones

1:00:08 that'll help speed up that testing so you can deliver you know new services based on these uh new features quicker

1:00:16 all right and we are at the top of the hour i'll close out with one last question um

1:00:21 uh infinera teresa i'll um i'll let you close out uh the q a with this one um

1:00:28 because i don't think i let you comment on this area so much when uh when uh when we were kind of wrapping up

1:00:34 um why is the management of coherent pluggable any different from um

1:00:39 the current optical management again some of that may have been addressed but but there is a question in and if you

1:00:44 want to add uh to that discussion um i think you know you thank you that's

1:00:50 actually that's a question where it's hard to be short but i'll try um so if you're if

1:00:57 you're um using a traditional optical network with transponders your transponders will have a very rich

1:01:03 interface that's specific and dedicated to optical management and functionality

1:01:09 with plugables your management is typically done with protocols like cmis

1:01:15 which are registered based and and very narrow in scope even though they're evolving um so from that already from

1:01:22 that point of view the functionalities that you have accessible are different and we you don't want that your

1:01:29 management is is actually blocking the evolution of functionalities that your optical pluggable more and more

1:01:35 sophisticated can deliver um so that's one concern the other concern is

1:01:41 if you go through the router to access the pluggable typical you you'll interface with a ip

1:01:49 controller while usually in an optical network you have an optical controller um that also introduces um some

1:01:56 subtleties in the management and and maybe impacts in mode of operation but i think actually that there's a great um

1:02:05 a specification that was published yesterday just by chance by the open xr

1:02:10 forum on a management

1:02:16 approach to coherent optical probables that complements the traditional

1:02:22 management protocols i invite everyone to go and check it it's worse to be read

1:02:28 i did see that out i haven't read it yet um i was gonna to make a comment and you

1:02:33 you kind of let into it that you coherent optics one of the trends we're following is it's mo of course it's starting the long haul it's moving more

1:02:39 and more closer to the user over time xr optics is one example of that but but

1:02:44 there are other areas as well and it's a big area of research for us is how far coherence is going to go at what data

1:02:50 rate um with that we are past the hour i appreciate everybody for tuning in uh

1:02:56 thanks for your great questions uh we will follow up with the white paper and please look out for it when it's

1:03:01 completed in july which has a lot more detail and of course thank you to our sponsors and to our each of the speakers

1:03:08 today for your insights and of course for helping out this entire project thanks everybody

1:03:14 thank you

1:03:52 you