Sterling Perrin, Industry Analyst and Consultant, LightReading

The Evolution of the Multi-Service Network Edge

AI & ML400G & 800G
Sterling Perrin Headshot
Still image with image of person in the upper left hand side, next to a slide being presented titled ‘Demands for high density multi-service edge routers.’

Service revenue is minted at the network edge.

The Multi-Service Network Edge has always been crucial to deliver on superior end-user experience as well as new, innovative service offerings. But, as in many technology areas, requirements keep evolving: Cloud-centric architectures, the overwhelming need for automation, machine-learning workloads, enforcing security policies, the insatiable thirst for applications for more bandwidth. That is just to name a few. 

Show more

You’ll learn

  • The key challenges for CSPs today and rapid changes taking place

  • Why there is an overwhelming need for compact routing platforms which support advanced services at scale

  • Advantages the emerging multi-service edge offers over traditional core and access routing technology

Who is this for?

Business Leaders Network Professionals


Sterling Perrin Headshot
Sterling Perrin
Industry Analyst and Consultant, LightReading

Guest speakers

Paul Liesenberg Headshot
Paul Liesenberg
Senior Product Marketing Manager, Juniper Networks
Joe Truman Headshot
Truman Joe
Senior Product Line Manager, Juniper Networks
Nikhil Rao
Senior Product Line Manager, Juniper Networks


0:01 and thank you for attending today's webinar we have evolution of multi-service network ad sponsored by

0:07 Jennifer before we begin I will cover a few housekeeping items on the left hand

0:13 side of your screen is the Q a if you have any questions during the webcast please type your question into the Q a

0:19 box and submit your questions to our speakers or questions will be seen so if

0:25 you don't get to answer you you may follow up your email at the bottom of your audience Castle are multiple

0:30 application widgets you can use if you have any technical difficulties please click on the yellow help widget here you

0:37 can find answers to common questions a copy of today's slide back is available for download in the green

0:44 resource list video near the end of today's presentation please take one minute to complete the survey that's

0:49 open on your screen your feedback is extremely helpful an on-demand version of that will be available about one day

0:57 after the event and can be accessed using the same audience link that was sent to you earlier today

1:02 I would now like to turn the event over to heavy readings senior principal analyst Optical and networks and

1:09 transport Sterling parent Sterling all right thank you Barbara and hello welcome everybody to today's webinar on

1:16 the evolution of the multi-service network Edge uh sponsored by Juniper

1:21 Networks um I'll be the moderator for today um but the

1:27 the heavy lifting for the speaking is going to go to our three uh presenters

1:32 today shown here on the slide um Paul lisenberg Senior product marketing manager with juniper uh Truman

1:40 Joe senior product line manager and then nikhil Rao senior product line manager

1:46 as well of course all all with juniper so they're going to trade off the duties

1:51 of presenting the material that we'll be going through and the flow for the webinar looks like

1:58 this uh kicking off with major market trends then getting into the new Metro

2:04 Edge in terms of both requirements for the network as well as some use cases

2:10 and then moving to the hyperscaler requirements as well as some use cases

2:15 around the hyperscaler deployments and then uh for the final section looking at

2:21 400 gig plus adoption which is is definitely a Hot Topic and something I'm

2:27 quite close to these days myself and then we'll wrap up and move to q a as

2:32 Barbara indicated please do ask questions throughout electronically we'll gather them up get to as many as

2:39 we can we typically try to have about 10 minutes at the end for Q a so please do ask the questions it's always an

2:45 important part of the webinar and the other place where the audience will come in is we have two poll questions

2:52 um looks like a little bit later in the presentation but there will be two poll questions so we'd be curious to get your thoughts and we'll share out those

2:59 results in real time uh when we get to those uh with that let me hand things

3:04 over to um I believe Paul you are going to kick this off

3:10 indeed I will start late thank you very much for the intro and hi everybody and thanks for

3:16 attending I'm Paul liesenberg with juniper and of course I'm going to be assisted by Dr Truman Joe and nikhil Rao

3:22 who will present their own sections as subject matter experts I'll just do a brief intro and then we'll move to the

3:30 different use cases the network Edge what we call the multi-service edge in Juniper it keeps

3:36 changing hence it needs to to always be ready to evolve and to adapt to even

3:42 more change and that is critically important because at the end of the day it's a it is at the multi-service edge

3:48 when you Innovative Services need to be delivered in the network and at scale

3:54 and furthermore of course that is where Revenue Service revenue is minted for

3:59 service providers so as I go through some of these Market forces some are Universal and they are

4:06 on everybody's mind these days basically and some others are quite specific to places in the network and use cases

4:13 we'll go through through both but let's start by revisiting some of the forces that are shaping markets right now like

4:20 continued exponential traffic we we have all heard this the rates of traffic

4:25 growth are still mind-boggling in general and primarily driven of course by the video and gaming at 4K and 8K but

4:33 there are other important traffic sources coming up like like machine learning and Innovative Network Edge

4:39 industry for all applications as well traffic growth in the end drives our need to implement faster and faster

4:46 interfaces not so long ago 10 gig was good enough now 100 Gig is becoming very

4:52 widely adopted and we're even looking into 400 gig and 800 gig right as you will see it later in the presentation

4:58 remember when one megabit DSL was fast asked enough things that have certainly

5:04 changed then there's a also the whole hybrid Workforce movement which greatly

5:09 influences of course Network adaptive adaptability requirements and whatever your adoption model is as a business you

5:17 want your service provider infrastructure to to support it based on whatever choices you make and that is a

5:24 critical role for the multi-service edge as well and we haven't called it out specifically in these boxes but I can

5:31 emphasize enough how much sustainability and power consumption become how important these have become for our

5:38 customers in some geographies more than others but this is something that we

5:43 need to address especially as we deliver more and more use cases closer and closer to the very edge of the network

5:50 with very high service density security is important we know that so the network has to provide and support ways to to

5:57 thwart the the sources of evil and what we call cloudification is also

6:03 important right or Cloud first it implies change traffic patterns but also

6:08 demands new Adoption patterns like especially stuff like programmability and cloudified leave spine architectures

6:15 and optimal support for very elastic and flexible Service delivery and also

6:20 operationally you look at Cloud best practices like orchestration Etc so that's making it into our world as well

6:27 quite dramatically and we could also we could also throw in

6:32 their supply chain challenges right they've been major it's serving a lot of priorities over the last 12 months and

6:38 it's pretty Universal so luckily it seems that things are easy and out there when it comes to that

6:47 so let's look at how those driving forces really reflect themselves and how

6:53 they impact our product strategy as well how does this impact the network Edge in the near future on the left you could

6:59 call it the the customer facing uh aspect of it you have traffic growth we

7:04 mentioned video at 4K and 8K new services and traffic patterns like like machine learning we did a very

7:11 interesting case study in collaboration with MIT just a few months back and but towards customers it is more traffic

7:18 especially more video and more subscribers and more devices I think the average person now has five to seven

7:24 devices here in the US and of course service providers need Innovation and they need to monetize new services and

7:31 Enterprises in turn need to provide even more complex Services internally at faster performance and with increased

7:38 security so on the right which you could call the core Network facing aspect of

7:43 it we have the demand for increased traffic and the faster speeds of calls of course but very crucial in these days

7:50 of cloud native architectures is we have huge peering demands to SAS as and is

7:57 infrastructures and we'll discuss those in a bit so in a niche nutshell we need

8:02 more service intelligence in the network and at ever faster speeds which these days tries for 100 Gig adoption and even

8:09 considerations for 800 gig readiness

8:17 so this diversity of requirements uh basically are different for the network

8:22 core and the network Edge but it especially reflects itself 100 in our

8:27 latest uh 306 ASAP capabilities for the multi-service edge and that's the MX product line mostly and I'll keep this

8:34 short because we really want to get into the use cases but we we firmly believe that the required service scale at the

8:42 multi-service edge is still served best by a specialized news generation silicon and I should note that we we don't do

8:49 this because we're into technology religion religion because developing silicon is expensive and and risky

8:56 especially now in the nanometer age but then again it allows us to deliver on

9:01 deliver on key design considerations such as we knew it had to enable very compact

9:08 form factors because collocation requirements drive that very aggressively and at the same time there's a need for power efficiency for

9:15 the same reason driven by energy costs in general and emerging requirements as well I just heard from one of our

9:22 customers in fact that they publicly stated they expect energy costs to search by 50 million dollars in 2023 for

9:30 their network operations which we can all agree is not insignificant and we need to make systems more sustainable

9:36 because of those forces then their service intelligence scalability they remain an overwhelming

9:42 requirement as well and we'll get into those with the use case discussions and

9:48 adaptability and programmability programmability are key areas that need to be enabled and for these Platforms in

9:55 the 400 gig and and even beyond the era the power of the 306 ultimately means

10:01 that they represent a Swiss pocket knife for many existing and emerging use cases

10:14 I introduced the 306 Asic now let me very briefly talk about the products

10:19 that the 306 Asic Powers the 306 is to support high density

10:24 service intelligence built-in security high speeds and very compact form factors and earlier this year we

10:31 introduced the mx1004 wait I'm on the wrong slide here

10:36 we introduced on the mx1004 and mx-304 platforms which are poster children for

10:43 design priorities that I mentioned they have millions of queues for a hierarchical qos check a very large fib

10:50 and red table support as well as tunneling and protocol encapsulation at 400 gig check as well the key thing is

10:58 that these platforms are ready for the future and and then you're emerging use cases but let me let our product management

11:05 experts get into the use case details we invited Dr Truman Joe who leads our mx-10k family as a resident expert on

11:13 emerging service providers use cases and take us through those and then nikhil Rao our product lead for the new mx-304

11:21 will then dive into emerging Enterprise government and data center interconnect

11:28 use cases let's start with Truman what dfrs Truman great thank you Paul

11:35 I'd like to start by discussing the evolution of the service provider subscriber access environment this is a

11:40 topic which in and of itself could fill many hours of webinars it's a complex discussion based on specific Regulatory

11:47 and business requirements we'll go over where we are and where certifieds want to go but let me note

11:53 excuse me that this is an open discussion you can follow in many industry forums that Juniper

11:59 participates in before we start though I think it's important to quickly introduce the

12:05 mx1004 which is not just for service provider subscriber delivery but is also

12:11 well suited for many Network use cases due to its modular nature compact form factor High Energy Efficiency and

12:18 scalable Network operating system the MX 1004 is part of the mx-10 000

12:23 series of universal Edge routers from Juniper Networks it reuses chassis components such as riding engines and

12:29 power supplies from the eight slot MX 1008 along with Lion cards such as the LC 480 optimized for one gig and 10 gig

12:37 interfaces and the lc9600 which is optimized for 100 Gig and 400 gig interfaces

12:43 its compact size makes it ideal for distributed scale out deployments or high throughput with a small blast

12:49 radius is desired such as in subscriber access environments

12:54 key features include 400 gig interfaces with Max tech support support for four line card slots and redundant critical

13:01 chassis components such as routing engines fabric cards power supplies and fans for high reliability High

13:08 availability as is true for all MX platforms it runs the genosis network operating system

13:16 um now let's get back to discussing the subscriber access environments you probably will not be surprised to

13:22 hear a fixed Broadband is driving digital connectivity with billions of consumers and Enterprises globally to

13:28 the tune of 1.2 billion in the last year this is a major Revenue generator for service providers but at the same time

13:35 providers at the same time provides a challenge due to the exponential growth

13:41 and changing traffic patterns these concerns can be addressed by bringing Services closer to the network

13:47 Edge and protecting it with embedded security space constraints found in these

13:52 environments mean deploying smaller form factor devices with low power consumption and high throughput

13:59 the drivers of the growth come from the cloud and software as a service allows growth of bandwidth intensive

14:05 applications such as 4K and virgin 8K video and gaming analysts estimate that 80 of traffic

14:12 today is video this is expected to grow at a 55 percent uh cumulative annual growth rate through

14:19 2030. at the same time stringent latency and performance requirements

14:26 excuse me from applications such as augmented reality virtual reality

14:31 Internet of Things uh Industrial Automation for example motivated service

14:36 providers to position user plane resources closer to the users to avoid backhauling traffic to Central

14:43 Beauty Pop in addition providers would like to simplify their operations for cost

14:49 reasons the traditional model large vertically

14:54 integrated platforms with massive internal fabric are used to support hundreds of thousands of subscribers

14:59 in many cases this creates bottlenecks that increase complexity and costs of engineering to maintain the network

15:06 what is needed is a more modular approach to BNG offering the simplified configuration software releases and

15:12 maintenance this has resulted in the development of the Broadband form of technical report 459 which defines BNG cups

15:21 music cups is Aggregates a traditional BNG architecture separating user and control planes and allowing each to be

15:28 independently engineered and scaled this disaggregation makes sense from a purely technical standpoint

15:34 as usual plane functions have more stringent memory throughput and latency requirements

15:40 architecture that lets you position BNG user planes anywhere in the network gives you flexibility to meet changing

15:47 custom requirements in more economical ways instead of always relying on huge centralized service Fabrics you have the

15:54 option to use smaller less expensive Edge platforms and scale out incrementally with demand

16:01 Juniper do compact adaptable platforms like the mx1004 and the mx-304 designed

16:07 for this new Service delivery world the use case shown here comes from a

16:13 Broadband provider in the Australian New Zealand region they wanted to reduce their total cost of ownership while

16:20 scaling services and maintaining best of class customer experience the solution was to disaggregate the BNG

16:26 using cups the control plane was cloudified and the user plane using MX devices were distributed using a spine

16:33 Leaf architecture the end result was a reduction of the blast radius from approximately 128 000

16:39 to 24 000 subscribers per device and at the same time the customer realized the

16:46 30 reduction in the TCO

16:53 and we have another use case for a North American Service writer for business services

16:59 the brighter was delivering High scale and reliable business services using a multitude of devices using a different

17:05 device for each service offering the issue they had was during high cost

17:10 and power space and other operational requirements of the network the solution was to implement a scale

17:17 out architecture using a compact platform that could provide multi-surge Edge functions with high throughput and

17:23 service scale to lower power and space requirements operational efficiency was achieved

17:30 using end-to-end control signaling with bgp which provided common control mechanisms independent of the service

17:37 type in addition support for advanced users enables seamless migration from bgp vpls

17:44 to evpn and from ldp RSVP based transport to segment routing

17:50 providers have gone through such architectural Transformations similar to these use cases

17:55 many times before and we know that these take time as there are operational challenges during the transition periods

18:03 General magazine transitions easy for our customers as we support all adoption and migration models from the now to the

18:10 Future Vision allowing them to eloquently coexist hence as we have done before we support

18:16 these architecture transition models that benefit our customers with best in case Best in Class

18:23 Solutions um so now let's look at emerging use cases

18:29 outside of the service providers subscriber access and we do have our

18:34 resin expert for many of those here in the form of nikhil our expert for large

18:39 Enterprise government and hyperscaler requirements the field

18:45 yeah hey hi thanks Truman uh hi all this is nikhil Rao I'm the PLM for MX 304 and

18:52 I mean this is great to be part of this session um so what we have been doing is constantly engaging and listening to our

18:58 customers the trends have been same I mean you've heard my friends Pablo and Truman speak of it in terms of what

19:05 should we be focusing on WE the solutions should be optimized for cost space and power these things have not

19:12 changed along with that what the expectations have been is that there

19:18 should not be any compromise on scale or performance now adding to this is security embedded

19:24 you want Security in terms of maxac or line rate encryption and also security

19:29 embed in the devices to make sure that when these devices are deployed in locations which are not owned by the

19:36 customers then the device should be secure I'm not tampering with the device should happen

19:42 and next is Rich Telemetry I mean the amount of streaming Telemetry that

19:48 device should offer should be able to troubleshoot the device or understand what is going on in the device

19:54 also future proof I mean the 400 gig is a real requirement and most of the customers are looking to Future proof

20:00 themselves with 400 gig and last but not the least is uh ready to deploy today customers want

20:07 something that is available with Rich feature set and I think uh MX 304 hits

20:13 all these boxes and is a huge success already we see huge demand from service provider cloud and Enterprise segments

20:21 for this box uh with the 4.8 T and also 3.2 t with

20:26 the Redundant control plane I think we we have multiple use cases already covered so let's uh let's dive into some of the

20:33 specific use cases that I've been working on uh one of the most complex use cases is

20:40 the virtual private Cloud use case uh we all understand and we have been

20:46 discussing this uh in this session and also previously from a very long time hyperscalar growth is driven mainly by

20:53 work for workload migration from on-prem to Cloud but what do you see going forward is some of the new things that

21:01 cloud is trying that'll actually increase the traffic three-fold or four-fold uh services such

21:08 as van cloud services that use that leverage the backbone of the cloud for

21:14 connecting different office locations I mean you already are connected to the cloud now you could actually leverage

21:20 those connections to consider the cloud as your van backbone I mean this will

21:26 mean there will be more traffic there will be more traffic passing through the cloud entering and exiting it and also

21:32 we have Enterprise 5G use cases so the amount of traffic entering and exiting cloud is going to increase so the focus

21:40 on how we monetize this and how we help our large hyperscale customers to

21:46 achieve this in the best way is what we're trying to talk about today I uh today the large hyperscale hours

21:53 already have hundreds of mxs in their network uh it is critical for them to provide direct connectivity to their

21:59 large customers at high speed in a reliable and cost effective way the focus has been to deliver services in a

22:06 reliable and cost effective way that's what differentiates this services from the traditional services

22:13 for this specific use case of VPC uh Edge let me break it down further VPC

22:18 Edge the virtual private cloud is a specific role that is used to terminate

22:24 many customer circuits at very high logical scale at the periphery of the network this is where the customer

22:29 enters the cloud network if you if you could say right I mean in this diagram on the right you can see that the CPE

22:36 devices connect to the VPC Edge this is where you hand off from the customer to

22:41 the network of the cloud and there are many ways of solving this you could have a meshed network of

22:48 connecting all the devices together but that would be extremely expensive so what has been working and what where MX

22:54 has made a difference is enabling cost effective ways of doing it a large tunneling and queuing scale of MX

23:00 combined with already the cost and power profiles and feature set is driving some

23:05 of these designs let's look at what design options were provided to our Cloud customers to solve

23:11 this problems with MX uh basically the peripheral Edge that's the VPC Edge could be a high scale Edge

23:19 router that processes traffic in a customer specific VPN context right forwards the traffic between the data

23:25 center and the customer Network this design is already being deployed by many and actually we've seen one of the

23:32 hyperscalar designs today the VPC Edge that is you call it vpch VPC Gateway or

23:37 peripheral device is already aware of the specific virtual Network location and hence sensor traffic towards an

23:44 intermediate Data Center uh Gateway which forwards the traffic to the specific virtual machine this is

23:50 already being deployed one another way of doing it is that the virtual machine location information itself is present

23:56 in the VPC VPC Gateway and no intermediate devices required this way VPC Gateway is uh

24:04 VPC Gateway has full information of virtual information of the virtual machine and what encapsulation is

24:11 supported and this can be used to directly connect to the directly connect and provide traffic to the customers

24:18 PPC Gateway today uh needs to understand the exact same encapsulation that the

24:25 server that the server provides thus eliminating the need of intermittent devices other than the features and complexity

24:32 we spoke of what is most important is the cost per customer and cost per bandwidth and also power consumption

24:38 power consumption and Cooling and cooling per device and per rack is

24:43 imposed by The Exchange facilities in which the locations are provided not all the vpch locations are owned by our

24:50 hyperscalers they could be positioned in pop locations which are owned by colos of the world

24:57 uh the focus has been and will continue to be under liability and cost and

25:03 that's why I think MX 304 hits most of those points to maybe to summarize where we feel that

25:10 MX 304 has made a difference in this use case is that many customers today are looking to

25:16 deploy 400 gig to for customer facing circuits with full Max Act and

25:21 increasing availability of 400 gig is accelerating this transition so MX 304

25:27 with feature and cap tunneling features cost inline Max act power profile seems

25:33 to be a very natural fit for this so we do see that this transition

25:39 picking up pretty well as well and second is something interesting design choice that we did with the

25:46 device that allows us to configure the mx-304 1.63 1.6 t or 3.2 with controlled

25:53 by redundancy of 4.8 T so that means that you could use the same device in different shapes and sizes

25:59 you could that will bring down the cost of sparing onboarding the device testing

26:05 and certification so overall lower TCO as well uh so along with our already proven

26:10 reach feature set and Compact and power optimized Solutions I think this this is actually going to fit in pretty well for

26:17 these use cases so moving on to

26:24 the next use case sorry I'll on the slide to uh Enterprise

26:31 uh use case right I mean now when we speak of Enterprise and the cloud

26:37 they're pretty much connected because I think most of the Enterprises are moving their workloads to the cloud

26:42 now Enterprise Solutions are getting more and more complex you have multi-cloud high bike hybrid cloud and

26:48 increased work from options that are posing challenges to how Enterprises have to think of their Solutions

26:54 and what is relevant today may not be relevant in maybe next one or two years itself right uh we already have

27:01 challenges with over substant van links we have cyber attacks to so leading to

27:08 security and strong DDOS protections and this is one of the number one things in mind for all the Enterprise cios

27:15 uh and I think end user experience

27:21 is extremely important for the Enterprise customers right I mean Enterprises are thinking of themselves

27:26 as service providers with the end users being their customers so the way Enterprises are now considering the

27:32 network designs is different from the past and also Enterprises must build networks

27:38 that caters to exponential bandwidth growth over the next five to ten years so uh in this slide I want to focus on

27:45 uh some of the complex use cases that is internet Edge but before going into internet Edge Let's see what are the

27:52 different van Solutions uh that are already there today and the how MX has been already deployed there right I mean

27:59 we already have the Enterprise van aggregation and backbone this is basically interconnection of multiple

28:05 types of Enterprise locations I mean you can have your branch headquarters and data centers connected that's the

28:11 Enterprise one aggregation and backbone uh internet Edge the internet Edge is an

28:16 interesting use case where I think interconnection of Enterprise Vans to one or more service providers allowing

28:23 access to users uh to the internet and also external corporate resources this

28:29 is getting more and more uh complex and diverging because of the way uh internet

28:35 is consumed today this clubs into the next use case that is the Enterprise multi-cloud the

28:40 interconnection between applications and micro Services across different Cloud platforms is actually adding a lot of

28:48 pressure on how do you design the internet Edge as well as the Enterprise multi-cloud connections right and last

28:53 is the data center interconnect I mean data center interconnect is a connection between Enterprise data centers that

29:00 enables resiliency like your availability zones if you could say so of all the four use cases that I spoke

29:06 of Enterprise one internet Edge Enterprise multi cloud and DCI we have

29:11 already mxs deployed today and uh maybe focusing on the internet Edge

29:18 uh what we are seeing is that the internet Edge is evolving

29:24 because it is not just used for backhauling the traffic uh from to the company headquarters to enable Security

29:30 Services like URL filtering anti-spam and IDF but also

29:36 considering that for now using it to connect to the cloud right I mean how how do you consider enabling your

29:43 existing infrastructure to work with multiple cloud and multiple internet

29:48 service providers uh what what is most important now is to

29:54 enable services like extremely high load balancing and filtering for DDOS protections so having

30:01 this Edge devices to be able to support filtering and uh sometimes encryption like maxac is is a

30:10 high requirement today and I think the ability to start small

30:15 and having different form factors was also helping this box I mean this solution

30:21 so basically if you take a step back and consider uh what Enterprise solution should offer is

30:27 improved operational efficiency increased flexibility and value of investment security and Cloud grade

30:33 reliability and I think uh 304 and even 1004 that way is proving to be a great

30:39 fit for these use cases we already have MX proven with Rich feature set and date

30:44 of protections with high filtering granular policy and telemetry also is encryption support and moving

30:52 from 10 gig connections to 100 Gig with future proof of 400 gig something that we feel will help our customers to move

30:59 forward with the journey so yeah with these two use cases I hand it back to Pablo Paul to take us forward thank you

31:07 all hey actually um I believe we have a poll question

31:13 next so I will I I will run us through the poll and gather up um

31:21 gather up the audience responses and push them out and um let you guys comment uh on the results this one is um

31:29 about the different use cases um between Truman uh and and

31:36 um and the kid we've gone through a number of use cases so this is um a set of them that the hybrid uh so this is

31:43 which is most interesting to you I believe it's it's multi-choice it should be um hybrid multi-service Edge

31:49 uh the BNG um control user playing separation cups migration the hyperscale data center

31:58 deployments or the 400 gig support and 800 gig Readiness so between the two

32:04 speakers um they covered that full set give you a

32:09 moment to pick um the ones that suits you and I'll push them out just waiting to get kind of a

32:15 critical mass here while I do that I will remind just looking at the clock we are definitely going to have a lot of

32:20 time for Q a which is great um if you do have questions please start

32:26 asking them now um it's always a little easier than waiting until the Q a begins

32:34 um but ask away now because we are going to have time for that let me push

32:39 the results here and uh this should show on everybody's screen the hybrid multi-service Edge 53

32:46 percent followed by about 26 percent the um

32:52 uh that 400 gig and 800 gig Readiness and then a little bit behind that

32:58 actually fair amount behind that the hyperscale interesting that the BNG got the least amount of traction let me just

33:04 do one more question I'll let um um I'll let I guess both Truman and

33:11 nikhil if you if you want to comment um Truman start with you and then Michaela if you have additional comments you want

33:17 to make but uh any surprises is this how you expected the results to come in Truman

33:24 uh I'm actually a little bit surprised I I thought that there would be a bit more

33:29 traction in the uh BNG control usual plane separation migration because in terms of our

33:38 customer interactions we certainly see a lot of interest there um but but perhaps uh perhaps given the

33:47 the topic of this webinar it might have skewed it a little bit towards the first first category

33:53 um um but yeah so I I am a little bit surprised yeah that actually is is a

33:59 good comment if you look at the title um we do have kind of you get it self-selects the subset of people so I

34:06 mean I guess it's a good thing the Right audience is is tuning in because that hybrid multi-service Edge is the top one

34:12 but you write it this may not reflect the overall population of operators globally

34:17 um Michaela any thoughts on on the ones related to uh to your use cases

34:22 yeah I think maybe Truman mentioned it well I I would have also thought BNG and maybe hyperscale are data center

34:29 deployments would have been higher but uh well this is not a surprise

34:34 considering that how multi-services Edge is evolving right I mean how do you want to see the future how do

34:41 you want to probably differentiate and yeah I think uh missile it is based on

34:47 the audience and it's a set that we have a snapshot of who we have great let me I

34:54 believe we've got another slide of content and then another poll

35:00 question and then we'll get into the Q a I actually am not quite sure who was taking this one was it um

35:07 I'll pick it over and yeah it wasn't interesting Paul as well I would have expected also that hyperscale would have

35:14 ranked quite higher but yeah I guess like you said the titles of the weapon are probably cell select

35:21 so thanks everybody and let's wrap this up briefly before we go into the Q a and

35:27 I hope this answered questions about the future of the multi-service edge that is growing in importance that's for sure it

35:33 is increasingly strategic to deliver on new services and monetization of course and commercial plugin here the MX family

35:42 is now enriched with the capabilities of the 306 here and now it's highly

35:48 adaptable we can't entirely foretell the future of course but we're ready for it

35:53 through programmability Power efficiency and with these new compact factors that allow us to deploy these powerful

35:59 devices pretty much anywhere so we're ready for things coming our way one

36:05 thing I want to mention you see here we mentioned 400 gig optimized and 800 gig

36:10 ready so we touched on this a couple of times and that is a great discussion the doctors also deserves a future webinar

36:16 all by itself just like PNG cups but the few words we we do indeed have customers

36:21 that have already taken advantage of our 400 gig interfaces of course we do that

36:27 and that said for an important part of our customers 400 gig at the very

36:32 Network Edge like here is where they want to be ready to go soon but is there a particular business needs require and

36:40 whenever they require it not all customers yet but it is a key strategic consideration for future proofing their

36:47 Network Investments and let's do 800 gig we can give away any roadmap information but it's

36:54 definitely on the technology Horizon and we'll see adoption patterns very much

37:00 like the ones of previous generations right where architectures are ready for it whenever the use cases we discuss

37:06 demanded and the key message is simple again 800 gig ready but I think the next poll could be interesting as well if we

37:13 can move on to that one all right yep let's

37:18 yes so our last poll question is on uh the 400 gig when do you expect 400 gig

37:25 uh deployments um and I mean I guess that's really the

37:31 question now where do you expect when do you expect uh the 400 gig deployments

37:36 um now uh 16 to 18 month windows so six to you

37:42 know a year and a half out or uh 19 to 36 month window so

37:47 um the next three years I guess if you don't expect them at all

37:54 um you probably put the well probably don't don't answer we probably should have put a a you know a further window but three

38:02 years is quite a bit of time so of those three pick what suits you best um

38:08 based on if you're a service provider for your own or if you're a vendor or someone else in the ecosystem

38:14 um serving the service providers you know based on your your own interactions with customers we'll give a minute

38:21 um for those results to come in and we'll do a push uh let me just push here and see what it

38:28 looks like uh let me do one more so interesting uh we do one last push and

38:35 see if we changed it doesn't look like it um yeah again it's a it's a it's a slice

38:41 of the audience we have um in terms of who we have on none saying right now six and then pretty

38:47 well about it is even split as you get uh the of the two options 50 50 60 to 18

38:52 and another 50 19 or 36 months out let me do one last push and see if it changes it is not uh Paul since you just

39:00 presented uh maybe let you weigh in first and if the others want to comment as well yeah I find it interesting that nobody

39:06 said now I would have expected at least one person to to vote for now and yeah I

39:12 think these adoption patterns are pretty typical if you look back at the doctrine patterns like over the history of

39:19 interfaces from 10 gig 40 gig 100 Gig 400 gig it's typically like this time

39:24 span that you have where where first of all like very large customers Mega data centers and very

39:30 large service providers Drive the business case but there's few deployments and then over time the

39:36 entire Market catches up and of course it's also a pricing question at the end of the day where initially these devices

39:42 are going to cost the cost report is going to be much higher and then over time the cost per pole actually

39:47 decreases and more people adopted

39:52 all right yeah any uh Truman or nikhil I don't know if you had any comments on this

39:58 if not um yeah I know just a quick comment I mean I I'm I'm a little

40:05 surprised that nobody said now um I mean there there are deployments we

40:11 know of that are going on in foreign

40:17 but otherwise you know the the results uh even breakdown between the upcoming

40:23 18 months and 36 months I mean it does take some time to roll out new technologies and customers do tend to

40:30 roll these out at different time periods so we we see that kind of behavior so I'm not that surprised that there's even

40:36 break down there I'm just surprised that there's nobody saying it now yeah the

40:42 only thing I would say on that is um it would be reasonable to assume the

40:47 hyperscalers the big ones are driving the initial deployments um our audience probably wouldn't

40:52 represent them too well we tend to draw heavily from the Telecom so this is kind of you know traditional Telecom of large

40:59 size but more of a Telco audience than uh you know the Googles of the world so that makes sense I think yeah I think

41:06 adding to what Truman said right I mean the VV are aware of deployments that are like either now

41:12 with the boxes that we have shipping that's being planned and already tested so yeah maybe uh and another thing is

41:19 right I mean this is not just core right I mean you we are seeing uh four and gig

41:25 to the core but what we are seeing with some of the new boxes is that phone only towards the customer that's something

41:31 that is pretty interesting right I mean we spoke of that one of the use cases for the virtual private Cloud segment so

41:38 the core is something that we all understand for forward gig to the core but what's what's interesting and what

41:44 is driving more interest is some customers asking 400 gig native with maxac capabilities turned on so that's

41:51 the kind of traffic growth that we are seeing so yeah interesting all right let us

41:57 um I believe we move right into the Q a from here uh we do have some questions so I'll

42:04 jump in and um again for the audience if you've got any additional questions

42:10 uh let us know now um Truman uh start with you and then of

42:17 course if somebody else wants to comment you may but the disaggregation the BNG uh cups um

42:25 use case application that you walk through a question about you know you

42:31 you addressed it a little bit actually in the poll question but maybe a bit more detail

42:36 um in the market traction that you are observing and any thoughts or recommendations that you have around

42:43 that um you know I don't know if you can name specific customers but types of customers regions any kind of detail on

42:50 on how you see that being deployed and where yeah I I can't obviously name specific

42:56 customers um agreements and all that but I will say

43:02 that there is actually very good Traction in the BNG cups

43:07 um much like the use case that I presented here there is actually a customer that's already deployed this we

43:14 have deployments basically worldwide for BNG cups uh quite a few are looking at

43:19 it qualifying it you know making that decision I I would expect

43:25 that there will be quite a few major customers moving in that direction

43:30 a good percentage of the market will end up there just because of the advantages that BNG cup provides the scalability

43:38 aspect the lower cost of ownership the ability to to leverage

43:47 the control plane separately from the user plane the user plan to be optimized in the hardware I think all of these are

43:54 are just um very good benefits of that model and I just see that from the from the cost

44:01 standpoint and simplicity of the of the model the ability to leverage smaller boxes which makes it a bit easier to

44:08 manage I see I see many customers moving in that direction and many have already moved in that direction

44:15 all right our dad something to happen it's everything that Truman said of course

44:21 but we should also keep in mind that any successful architectural migration it takes some time right and while you're

44:27 engaged in that change you will invariably have to support a hybrid transition architecture and I think it's

44:34 a great help for our customers that we do that very effectively because we cover every type of architecture out

44:40 there great uh there's a question around automation I think Paul maybe start with

44:47 you and again if anybody else wants to add to it um

44:53 it references the some comments made in the presentation around automation

44:58 um and they're looking for more detail in terms of what what you know what I guess what what are

45:03 the tools involved uh the software that's involved in Automation and and um

45:08 you know maybe just to add on that a little bit to where are you seeing the biggest traction with with automation

45:16 the automation is a huge discussion all by itself and we have done a few webinars that the products we're

45:23 discussing today fit under the mostly under the Paragon automation umbrella if you look at our product offerings and

45:30 but automation is also one of those things where you know different customers also have very different goals

45:35 and when it comes to configuration to management to security or troubleshooting so if in general I would

45:42 say if you're interested in automation check our Paragon page on the Juniper website and stay tuned for webinars

45:49 because it is a very hot topic indeed

45:54 did anyone else want to add to that question if not I'll go to the next

46:00 nope all right um question on um power consumption

46:08 um and I guess maybe two two parts to this one the question is very specific to the the power consumption

46:15 um on on the product lines that are being discussed and I would add to that

46:22 um to maybe even broaden this out a bit I'm curious how important power consumption

46:28 is in your discussions with with customers across both you know maybe for nikhil

46:33 and then also separately for for Truman in terms of the priority um just for a little bit of context I'm

46:38 in New York right now we did our 5G transport event um yesterday and

46:44 um one of the comments that came up that we didn't have a lot of discussion around power um uh consumption and kind of green and

46:51 sustainability and his comment was made that if the event had been held in Europe probably would have been a higher

46:56 priority um so I'm just curious I guess too that the specifics about the consumption on the products but two also

47:04 um how important that type of metric is and it doesn't vary by region just out

47:09 of curiosity uh maybe Truman start with you and then we'll go to nikhil oh sure thing

47:15 um to put it very simply I think power consumption is a very important attribute

47:21 um this is something that I find many many customers ask about how many

47:27 customers um have it very high on their priority list um and and I think rightfully so I think

47:35 um as as good stewards of the planet at one level you know we want to make sure

47:41 that products are sustainable that powers use as efficiently as possible

47:48 um and and there's a very real cost to it as well for for customers power is

47:53 not inexpensive and the cost of power is going up energy costs are going up this

48:00 results in a very real impact to operational expenses for a network

48:05 um depending on your size of your network the power bill can be in the hundreds of millions of dollars

48:12 um so so you know it's very important that there's no snow no way around it

48:18 almost everybody asks almost every RFP I've ever seen asks about it so it's

48:24 very very important in my mind I I think everyone does care and probably should care if they don't care they should care

48:32 um in terms of more specifics now I mean that one part of the question that that you mentioned starting was

48:38 um the power for for the 10K Price Line specifically that's a hard hard question to answer in

48:45 a general q a uh because the power consumption of course will depend on

48:51 the exact configuration the traffic you're running the environment you're running in

48:57 um so so please take what I say with a grain of salt uh and and I will just

49:04 claim it by saying your mileage will vary for sure depending on your environment but typically on a on an

49:10 mx1004 which which was the product that we had mentioned in this presentation uh

49:15 with a couple of line cards configured in sort of a typical traffic load let's

49:21 say 50 traffic running through the box at sort of an Internet mix I think you

49:27 can expect to see something in the order of three or four thousand watts of power being consumed by that box

49:33 um but but again can't stress enough your mileage will vary it really depends on configuration traffic load Etc so all

49:40 right thanks thanks for that and nikhil same kind of two two questions specific in the

49:47 general for your your product set I think when you pay directly to the power bills

49:54 or when you pay the co-location owners there's a there's a difference right I mean these boxes are going to color

50:00 locations and power and space is like the most important thing that we have right I mean obviously you need the

50:06 feature set but the amount of power savings translates to real dollars double the value basically so and

50:14 there's also restricts on some how much of power you can put in per rack right so how do you design your rack these all

50:22 not all locations have 10 15 000 Watts coming to the each full rack

50:27 so if you want to have consistent designs and if you are constrained by the total rack I mean power is important

50:33 every cloud customer I speak to First we start with the power number and then go back

50:40 that's how one of the design decisions are taken cost and power is extremely important so I mean I I think power is

50:48 like number one ask at this point from our hyperscale as yeah and in specific GEOS we are seeing

50:55 request more compared to some of the other GEOS as you mentioned right initially uh the event was in America North America and if it was an email it

51:02 would have been a different thing yes we see that uh very much and uh there is

51:07 more and more questions on what will be the exact

51:13 power at this the particular amount of traffic and all this location that's what Truman alluded to also I think the

51:20 question now is that we are seeing interest in specific customer deployments Also earlier you had a data

51:25 sheet that would tell you what was the power look like you put in add some plug Optics or maybe overall basic numbers

51:31 but now everyone wants to know specifically at my usage at my temperature at my altitude at my traffic

51:38 mix with without without Mac Sac how much of power it is so those scenarios have changed and uh we see more

51:45 questions in those areas and just to add I mean if you're looking for uh I mean working with hyperscalers

51:52 and if you're looking for a box that has control but redundancy that does uh 3.2 t with basic 50 percent Timex traffic

51:59 we're looking at 1000 200 Watts and uh

52:05 this is this is driven by again the one of the latest edition silicon right I mean because it's 7mm and it's power

52:11 efficient we're able to provide better efficiency in terms of power all right excellent I appreciate that

52:17 yeah and and good comments I expected um that even the hyperscalers you you

52:23 would say power would come in number one and I think it's a very accurate comment about the Colo I I hear that myself

52:28 quite a bit when it's moves to the Colo world and you're and you're paying for that power uh and it's very expensive in

52:34 these um data centers that it becomes you know sustainability environment is is of course important but there's also

52:41 some very um hard knowledge and sense that that come in uh with that we are pretty close to the hour uh we've had kind of a

52:46 lengthy q a session so let me close it out here um I want to thank all three of our

52:52 presenters today for your your comments your presentations and answering our questions and of course thanks to

52:58 Jennifer for sponsoring and thank you to the audience for tuning in we'll we'll close out here thank you everybody

53:03 was our pleasure thank you so much thanks everyone

Show more