Zeus Kerravala, Principal Analyst, ZK Research

Juniper Network Demonstrates Telco Cloud and Automation at MWC23

Industry VoicesDemo Drop Telco CloudNetwork Automation
Zeus Kerravala Headshot
Still image shows a photo of Nick Davey, Sr. Product Marketing Manager at Juniper Networks on the left side with the Juniper Networks logo underneath and a photo of Zeus Kerravala, Principal Analyst of ZK Research on the right.  In the upper right corner there isa symbol with the word ‘SHORTS’ under it.

In this video, Juniper’s Nick Davey demonstrates the company’s telco cloud and automation solutions. During the demo, Davey shows how network layers can be collapsed – from the data center underlay to cloud overlay. This enables cloud-to-DC connectivity to be simplified and automated. Davey demonstrates how to deploy a network fabric in minutes using Juniper Apstra.

Show more

You’ll learn

  • How network layers can be collapsed – from the data center underlay to cloud overlay

  • How a Kubernetes service can quickly and easily be connected to the fabric

Who is this for?

Network Professionals Business Leaders


Zeus Kerravala Headshot
Zeus Kerravala
Principal Analyst, ZK Research

Guest speakers

Nick Davey Headshot
Nick Davey
Sr. Product Manager, Juniper Networks 


0:04 yeah here now with the last demo in the

0:06 in the Juniper Ruth

0:08 um with Nick Davey Nick we want to give

0:10 a quick intro to yourself yeah hi my

0:12 name is Nick Davey I lead the product

0:13 management team for our Telco Cloud

0:14 Solutions and we will be talking Telco

0:17 cloud and automation which are two

0:19 things that really haven't gone hand in

0:20 hand ever historically right so what are

0:23 you gonna be showing us here uh what

0:24 we're going to be showing you is how we

0:26 can collapse the layers of networking

0:27 from the cloud overlay to the data

0:30 center underlay and we can automate

0:31 connectivity coming out of the cloud and

0:33 connect that to the physical Network

0:34 underneath it and the cloud actually has

0:37 a lot of you know dare I say Game

0:39 Changer potential for telcos right it's

0:42 something that they really haven't taken

0:43 advantage of historically that's right

0:45 and that's mainly because of the

0:46 complexity that comes with cloudifying

0:48 all of your apps right yeah and so the

0:50 big networks too yeah and big networks

0:52 operating at scale so the tools that

0:54 we've created allow to automate all of

0:56 those kind of complex points so we'll

0:58 take you through a couple scenarios

0:59 right now

1:01 so we're going to start off by showing

1:03 how we can Fabrics in minutes with our

1:05 intent-based abstra automation solution

1:09 and I want to capture is your intent

1:11 based Navigator engine that's actually

1:12 the only multi-vendor one in the

1:14 industry too yeah that's right yeah

1:16 and um what we're going to show you is

1:18 how we abstract away The Logical

1:20 configuration topology of a cloud

1:23 network from all of the vendor specifics

1:25 so what you're seeing here is we're

1:27 defining the layout the blueprint really

1:29 of the network we're going to create

1:31 we're going to give it all of its assets

1:33 and attributes and then once we're ready

1:35 once we have the staged we can roll it

1:37 out by applying it to a set of devices

1:40 now this is where the multi-vendor bit

1:43 comes in because we have our general

1:45 design by selecting the device profile

1:47 and the interface map we can say how

1:49 what Hardware we want to map that

1:51 configuration onto

1:55 once we have the configuration mapped we

1:57 can stage it up we can see all of the

1:59 changes we're going to make to our

2:00 Network we can review them we'll add a

2:02 commit message so we know what we did

2:03 and then we can apply that to the

2:05 network and it's validated before commit

2:07 yeah that's right because historically

2:09 people would commit things to networks

2:12 and then they wouldn't find those broken

2:13 until after they've committed it until

2:15 the screen test right and then they need

2:17 to roll it back and but this avoids all

2:19 of that yeah because with App Store what

2:21 you're seeing is we monitor the actual

2:23 roll out of the device you can see you

2:24 know that happy green semi-circle

2:27 rollout as all the configuration was

2:28 applied to the network yeah and all the

2:30 services come up like essentially what

2:32 this is doing it's creating a digital

2:33 twin right of the environment yeah that

2:35 you can run tests in and yeah yeah

2:37 exactly it's creating like a virtual

2:39 multi-vendor kind of or I should say a

2:41 device agnostic copy of the network

2:46 everything that we do is inherently

2:48 multi-vendor here because we start off

2:50 with these abstract topologies and

2:52 Abstract devices so when we're adding

2:54 devices and racks we're not really

2:56 interested in the configuration we're

2:58 interested in you know the layout the

3:00 properties of the networks you can see

3:02 this is another copy of that same

3:04 topology running Arista Hardware we

3:06 don't really make too many bones about

3:07 it in the configuration we can even use

3:10 Cisco and xos what matters is the

3:12 topology of the network now when we're

3:15 rolling out Services

3:16 evpmv excellent services are the same

3:19 regardless of the vendor right the whole

3:20 point is interoperability so we Define

3:22 the characteristics of the network we

3:24 want to create got it yeah

3:26 once we have the subnet the vni

3:29 automatically assigned we're ready to

3:31 push this to the network

3:35 and we'll see in hyper speed right now

3:37 we're going through and provisioning

3:39 this the cool part is we just have to

3:41 pick out the endpoints we want to

3:42 connect this to abstra is really

3:44 interested in just the endpoints and the

3:45 Canon activity so where do we land the

3:47 service

3:49 what's the beauty of the fabric right

3:51 it's sort of independent of the

3:53 underlying infrastructure yeah exactly

3:54 when we here I'll just pause my hyper

3:56 speed here if we actually look at the

3:58 config what apps just done is it's

4:00 rendered the configuration into the

4:01 device specific config uh before it

4:03 applies it to the network and we can

4:05 review that as well one of the beauties

4:07 of abstra is it's built by Network

4:08 engineers and so it doesn't do anything

4:10 to abstract important details like

4:12 configuration like configuration drift

4:14 you can actually see what's going on but

4:16 you don't need to bother yourself with

4:17 those details you can just apply the

4:19 config

4:20 [Music]

4:22 so we'll see we'll review that config we

4:24 can go in and again give ourselves a

4:26 happy commit message and then we can

4:27 commit that to the network

4:36 and it's all configured yeah and that's

4:38 it again now in our virtual data center

4:40 we have connectivity but just in the

4:43 interest of time I want to skip ahead

4:44 and show you how we can interconnect

4:46 something like a kubernetes service to

4:48 this fabric we just built

4:50 so we're going to show you how we can

4:53 add connectivity for an sraov virtual

4:55 function or a VLAN that needs to go to a

4:57 kubernetes pod via appstra this all

5:01 happens because we've created awareness

5:02 between the sdn layer and kubernetes and

5:04 the data center fabric orchestration in

5:07 order to do that we tell the the sdn

5:09 solution the location of the App Store

5:12 fabric and we give it some information

5:14 around the the blueprint it should

5:16 consume as well as the tag or the label

5:18 that we should apply to services that we

5:20 want connectivity for

5:22 once we've created that we have the

5:25 ability to map virtual networks inside

5:27 of kubernetes to the fabric itself

5:30 now this is important because uh

5:32 previously these two networks were kind

5:33 of ships in the night so whatever you

5:35 create inside of kubernetes the fabric

5:37 doesn't know about it connecting it is

5:39 hard and so now what we've done is we

5:42 have visibility inside of the actual

5:45 kubernetes networks for the underlay

5:48 and here's the the mandatory pile of

5:50 yaml so you can see it is indeed

5:52 kubernetes

5:57 once that Network's available we can

5:59 actually consume it by creating a pod

6:07 now the neat thing here like although

6:09 we've created our pod so we can see

6:11 we've got connectivity inside of our

6:13 kubernetes container

6:15 um just having connectivity isn't enough

6:17 yeah we need to show that we've got a

6:20 healthy connection as well to the fabric

6:22 so we can actually verify the state of

6:24 that configuration so we can see in the

6:26 network not only do we have our Network

6:27 provision but we see that in abstra it's

6:30 been provisioned as well so you have

6:31 that closed unit provisioning yeah and

6:33 that's a strength apps is the closed

6:35 loop is what lets it be intent based

6:37 yeah absolutely it's not about what

6:39 you've configured it's about what you've

6:41 got running right yeah and now by

6:43 creating this linkage between cn2 and

6:45 kubernetes and the data center fabric

6:47 managed by abstra you have that

6:49 connectivity that goes from the top of

6:50 the Telco Cloud all the way out to the

6:52 fabric okay got it

6:55 and that's really that I mean kubernetes

6:57 availability and visibility inside of

6:59 abstra fabric provisioning inside of

7:01 kubernetes it gives you that whole

7:03 service from tip to tail and automated

7:06 Telco Cloud experience yeah well Telco

7:08 Cloud's been something that's been

7:09 promised for a long time and you could

7:11 argue that without the automation piece

7:13 it's you know really not all that

7:14 beneficial so the expertise needs to

7:16 live somewhere it either needs to live

7:17 an operations teams that are running

7:19 this and through like a intricate

7:21 engineering or it can live in high level

7:23 tools that are built to solve these

7:25 problems and that seems like a much more

7:26 cost effective way to do it and this is

7:28 how we're going to scale it out yeah all

7:29 right well thanks Nick I appreciate it

7:30 that was good cool thank you very much

Show more