AI-Driven Campus Fabric Walkthrough

00:00 this is the overview of ai driven campus

00:03 fabrics

00:04 with the juniper mist cloud

00:06 these cloud native architectures are

00:08 based on three concepts

00:10 design deployment and operations this

00:13 overview will cover day 0 day 1 and day

00:16 2 aspects of campus fabric deployments

00:19 and how juniper mist has automated this

00:22 process

00:23 every ex switch shipping today has a

00:25 unique claim code and this code can be

00:28 used to automatically onboard devices

00:30 directly to the juniper miss cloud

00:33 you can also use the missed ai app to

00:35 easily claim switches by scanning the qr

00:38 code and immediately onboarding your

00:40 device

00:42 once the devices are installed and

00:44 powered up you will automatically see

00:46 them in the switch's dashboard

00:50 once the devices are onboarded you can

00:52 now deploy the campus fabric with no

00:54 console cables needed

00:56 here are the steps involved first choose

00:59 the topology you want and assign device

01:01 roles

01:02 juniper sports various topologies

01:05 evpn multi-homing which is a two-tier

01:08 deployment

01:10 campus fabric with core distribution

01:13 or ip close

01:14 with the last two being three tier

01:16 architectures

01:18 so after we choose the topology and

01:19 assign device roles

01:21 examples are collapse core access core

01:25 distribution with access

01:27 step two is to configure the networks of

01:29 interest

01:30 step three

01:31 define the physical connectivity step

01:34 four apply the configuration and that's

01:36 all that is required

01:38 let's take a look at a simple example of

01:41 a campus fabric deployment we are

01:43 looking at a multi-homing campus fabric

01:45 topology with two collapsed cores

01:47 the ex4400 devices and an access switch

01:51 which is an ex4300

01:53 our network has four vlans employee

01:56 contractor guest and iot

01:58 we want the employing contractor to be

02:01 internal and be able to talk to each

02:03 other and the guests and iot will be

02:05 vrfs of their own

02:07 let's look how we would configure evpn

02:11 the first step is selecting the topology

02:13 in the ui

02:14 evpn multi-homing in this case

02:18 now we select the devices the two 4400s

02:21 which are set up for ztp

02:23 we will add the

02:24 ex2300 as part of the access

02:28 now let's configure the user networks

02:30 and add the four vlans

02:32 employee contractor guest and iot

02:36 the gateways are on these core devices

02:38 and will provide either dhcp or static

02:41 ip addresses

02:43 now segmentation

02:46 we associate employee and contractor to

02:48 internal resources and the other two

02:51 will be vrfs of their own

02:53 we need to allow communication between

02:55 the axis and core switches and make sure

02:57 the four vlans are allowed the rest is

03:00 done automatically including the mtu

03:03 which is automatically set

03:05 the last part is how the devices are

03:07 physically connected

03:10 how are the cores connected to each

03:12 other

03:13 and how are the cores connected to the

03:15 access switches

03:17 we use esi lag

03:19 now just select two uplinks connected to

03:21 the core devices

03:24 we are done and we'll confirm

03:26 now all of our evpn changes are done and

03:29 saved to the cloud

03:31 now we look at marvis actions which is

03:34 really the morning coffee view of the

03:35 network

03:37 showing the important issues that you

03:39 the administrator should be looking at

03:42 across

03:44 1127 devices and 55 000 ports

03:48 marvis actions has detected 25 bad

03:50 cables showing the exact port and site

03:53 details two negotiation mismatches

03:56 four devices experiencing continuous

03:59 port flaps

04:00 we also show you persistently failing

04:02 clients

04:04 if a client is continuously failing

04:06 authentication it could slow down your

04:07 network and your radius server

04:10 marvin's actions is designed to show you

04:12 ai verified problems that you can either

04:15 solve or let marvis open a ticket for

04:17 you with all of the relevant information