Service Management and Orchestration Unlocking the True Potential of Open RAN

00:06 so we're here to talk about openran

00:07 this is a big topic with lots of

00:09 potential i'm here with uh constantine

00:10 from juniper networks

00:12 constantine great to see you why is

00:14 juniper

00:15 active in openran hi gabriele good to

00:17 see you too

00:18 um that's a good question well juniper

00:21 has been a leader in high performance

00:23 performance networking uh focusing on

00:26 delivering sla

00:27 in very complex networking environments

00:30 now in the world of 5g

00:31 now we have a new opportunity to really

00:33 stretch the sla

00:34 and the slos across the infrastructure

00:37 in a

00:38 immense infrastructure highly

00:40 distributed infrastructure

00:41 all the way through the airwaves to the

00:43 devices right

00:44 whether these are iot devices or

00:47 smartphones

00:48 smart cars whatever and oren

00:51 gives us that opportunity because with

00:54 the disaggregation the software driven

00:57 radio think of the radio intelligent

01:00 controller

01:01 as the operating system of the of the

01:03 run

01:05 this is a green field opportunity i

01:08 believe that

01:09 with the moves that juniper did with

01:11 acquisition of natsia perhaps the most

01:14 innovative solution of the radio

01:16 intelligent control in the market today

01:19 we are now in a strong position to

01:21 really bring

01:22 the innovations that juniper is known

01:25 for

01:26 across uh through the radio as i said

01:30 uh and deliver uh strong slash to 5g

01:33 devices

01:35 yeah i think that's a great point about

01:37 bringing across you know expertise and

01:39 knowledge from from

01:40 other domains where there's really been

01:41 a focus on performance and sla and

01:44 programmability determinism everything

01:46 like that

01:47 so this um this rick this ran

01:49 intelligent controller product

01:51 um that's a kind of key part of the oran

01:54 architecture what are you kind of

01:55 expecting that to

01:56 to deliver for for operators right

02:00 um i like to call the radio intelligent

02:03 controllers the operating system of the

02:05 new

02:05 radio disaggregated radio architecture

02:08 right and if we

02:09 think about it as such which i believe

02:12 it is

02:13 then we can start thinking about the

02:15 applications we can run

02:17 on the rig right essentially

02:22 oren specifies apis on the du side that

02:26 allows us to hook into the low level uh

02:30 mac layer scheduler and be able to

02:33 modulate

02:33 how resources over the spec over the air

02:36 are allocated to competing flows

02:39 competing users competing slices

02:42 and therefore using intelligence

02:45 real-time intelligence about

02:47 how users behave how the radio addresses

02:51 demand we can modulate resource

02:55 allocation

02:56 and deliver strong sla right so

02:59 juniper is known for its transport

03:01 networks actually

03:03 it foresaw the need to address some of

03:06 the use cases of 5g in particular

03:08 network slicing

03:10 and today juniper supports network

03:12 slicing across all of its transport

03:14 products right

03:15 so this was a natural evolution for us

03:17 to move into

03:18 a green field opportunity on the radio

03:21 and provide applications

03:22 that address everything from the

03:24 well-known zone

03:26 type of optimizations to

03:29 new capabilities in the radio that have

03:31 to do with

03:32 very fine grain smart admission control

03:37 network slice sla assurance

03:41 strong visibility you know that

03:45 allows us to hook mlai type of

03:48 applications

03:49 and provide opportunities not just to do

03:52 extreme optimization from the device

03:54 perspective from the network perspective

03:56 but also to open up opportunities for

03:59 our customers our operator customers

04:01 to monetize you know the the you know

04:04 investment they have in the radios by

04:06 offering new applications like spectrum

04:08 series

04:09 massive mime optimization and on and on

04:12 and juniper innovates and

04:16 leads in that space not just by uh

04:19 being you know the first in the in the

04:22 world of

04:23 radio access optimization through the

04:26 rig

04:26 but also in terms of some of the

04:29 applications that we're bringing to the

04:30 market

04:32 okay and these these applications these

04:34 x apps and our apps

04:35 that run on the ric platform are these

04:38 being

04:39 internally developed by juniper or you

04:41 look into partners or

04:43 you know different different put it this

04:46 way independent software vendors to come

04:48 and fulfill some of that

04:49 absolutely the success of warren depends

04:51 on

04:52 enabling operators to select the best of

04:56 breed

04:57 right so uh to that end

05:00 we have to support any type of x up or

05:03 our app

05:04 no matter where it comes from so x up

05:06 and r portability is absolutely

05:08 strategic to the success

05:10 of warren and we definitely support that

05:12 through our sdk

05:15 but we're also developing our own x apps

05:18 as well

05:20 which i think bring a significant

05:22 differentiation in the overall solution

05:24 we

05:24 uh you know we bring to the market no

05:28 yeah okay i wanted to move on to you

05:30 mentioned network slicing that's

05:32 obviously like a

05:33 key capability in 5g particularly we

05:35 think about enterprise services

05:38 um just uh to talk through a little bit

05:41 how

05:41 how the ric will help

05:45 kind of enable operators to offer

05:46 network slicing and over-the-air

05:48 uh kind of service guarantees and things

05:50 like that

05:52 right i like to use

05:55 an analogy that may sound a little

05:57 preposterous

05:58 when it comes to network slicing right

06:00 what is network slicing

06:02 it can be anything from the familiar

06:06 service training on one end of the

06:09 spectrum

06:10 to uh what i call an mvno

06:14 on top of a you know highly distributed

06:17 you know tier one operator including

06:19 public cloud so

06:21 my goal and our goal at juniper is to be

06:24 able to

06:25 programmatically declaratively bring up

06:27 an mvno

06:29 planet scale in a matter of a few hours

06:32 not months and years that's a slice

06:35 that's the other end of the spectrum

06:37 so slice covers pretty much anything

06:41 that you can think of right tenancy so

06:44 you need to have

06:45 separate uh you know different tenants

06:48 uh

06:49 you know with a strong sla per tenant in

06:51 terms of dimensioning

06:53 how many users each tenant what type of

06:56 connection

06:57 is that going to be an emv or a massive

07:01 machine to machine communication type of

07:05 slice or the familiar

07:08 ultra low latency high reliability

07:12 type of slice and everything in between

07:14 those uh

07:15 3gpp defined use cases right so a slice

07:19 can be anything and we need to address

07:20 segmentation

07:21 security and above all sla

07:25 and having the ability to do so in an

07:27 end-to-end

07:29 fashion uh is you know

07:32 uh makes the rick an absolutely

07:35 necessary component to be able to

07:36 stretch the sla to

07:38 deliver the sla over the spectrum um

07:41 so we're taking a very you know broad

07:44 view of

07:45 network slicing and as you're i'm sure

07:48 you know

07:49 uh network slicing has been deemed to be

07:51 the perhaps

07:52 the most important use case in 5g

07:57 and several analysts have attributed or

07:59 have attached

08:00 you know hundreds of billions of dollars

08:03 of monetization

08:05 that will come through the benefit of

08:07 network slicing networks

08:09 for the operators in particular

08:12 yeah yeah i mean um the we

08:15 saw a great example recently actually we

08:17 had a we had an event on openran and one

08:19 of the keynote speakers was from a

08:21 japanese operator

08:22 kddi and they showed a a proof of

08:25 concept they've been running with ram

08:27 slicing using uh

08:28 using a rick and i had a camera on um

08:32 on the front of a train running through

08:33 this really deep kind of urban canyon in

08:35 in tokyo with gantries and metal and

08:38 different moving trains on the one hand

08:40 they showed performance

08:42 you know just on best effort and it was

08:44 up and down every time you went past a

08:45 train came by the

08:47 the performance was dropping and

08:49 bursting all over the place

08:51 they ran the same line the same camera

08:53 on the same train with a with a rick

08:54 enabled

08:55 application and it was just a solid 30

08:57 megabits per second downlink you know

08:59 barely any variation just a proof of

09:02 concept but but a really interesting one

09:04 you

09:04 think what that could mean in in future

09:08 so in the o-ram we have this uh service

09:10 management and orchestration layer

09:12 um i guess what i'm interested in is how

09:14 when you talk about these slas and

09:16 and kind of supporting them end-to-end

09:19 how does the smo

09:20 tying with the wider network

09:22 orchestration picture

09:24 the smo is critical in being able to

09:27 realize to provision

09:28 a slice across across the three semantic

09:31 domains which is the radio the transport

09:33 and the core

09:34 and we have multiple incarnations of the

09:36 transport for example we have the

09:38 you know mid hall front hall backhoe

09:41 we can have multiple instances of the

09:43 core um

09:44 and of course the radio you know can be

09:47 realized

09:48 through the edge cloud a lot of the

09:49 radio components and we count there

09:52 probably in the

09:52 thousands maybe for tier one tens of

09:55 thousands of clouds you know it's clouds

09:57 that uh

09:58 host the duc etc so how do you bring

10:02 the benefit of network slicing across

10:04 such an immense infrastructure

10:06 and uh bring the locality aspect of it

10:09 because you may want to have a slice

10:11 that is local to a particular region or

10:13 it can be across the entire

10:14 infrastructure

10:16 you need to deliver the sla that ties

10:19 how the rick works with a transport to

10:22 ensure that

10:23 you know sla the latency and the

10:25 throughput are delivered across the

10:27 infrastructure

10:28 you need to be able to provision again

10:31 the core network functions to cope with

10:33 the traffic and the dimensionality of

10:35 the

10:36 of the slice so all of that comes

10:38 together

10:39 under the smo uh which is a fairly

10:42 complex problem

10:43 right and you know activity yeah across

10:47 yeah yeah fantastic i look forward to

10:48 seeing how this how this all evolves as

10:50 you know

10:50 as as juniper and your operator

10:53 customers and whole industry works

10:54 through

10:55 uh some of the challenges and puts this

10:57 to work um constantine with juniper

10:59 networks thanks very much