Download "Introduction to L2/L3 EVPN Services Supported on the Nokia 7750 Service Router"
%3Aformat(webp)%2Fi.ytimg.com%252Fvi%252F8sSMoQEkX4s%252Fmaxresdefault.jpg&w=1139&q=75)
Please wait. We're preparing links for easy ad-free video watching and downloading.
Similar videos from our catalog
%3Aformat(webp)%2Fi.ytimg.com%252Fvi%252FA1Rg3Y4F4Fs%252Fmaxresdefault.jpg&w=1139&q=75)
%3Aformat(webp)%2Fi.ytimg.com%252Fvi_webp%252FAEEq7GMolls%252Fmaxresdefault.webp&w=1139&q=75)
%3Aformat(webp)%2Fi.ytimg.com%252Fvi%252FKPteuQTxhRw%252Fmaxresdefault.jpg&w=1139&q=75)
%3Aformat(webp)%2Fi.ytimg.com%252Fvi%252FKb_kIq5tMGQ%252Fmaxresdefault.jpg&w=1139&q=75)
%3Aformat(webp)%2Fi.ytimg.com%252Fvi%252FMQ7hFY3EfpA%252Fmaxresdefault.jpg&w=1139&q=75)
%3Aformat(webp)%2Fi.ytimg.com%252Fvi_webp%252FYb-upYhy08o%252Fmaxresdefault.webp&w=1139&q=75)
%3Aformat(webp)%2Fi.ytimg.com%252Fvi%252FZj0_VFT0op0%252Fmaxresdefault.jpg&w=1139&q=75)
%3Aformat(webp)%2Fi.ytimg.com%252Fvi%252FoTtfbZT49GE%252Fmaxresdefault.jpg&w=1139&q=75)
%3Aformat(webp)%2Fi.ytimg.com%252Fvi%252FylhHYtTyVGE%252Fmaxresdefault.jpg&w=1139&q=75)
%3Aformat(webp)%2Fi.ytimg.com%252Fvi%252F6BobSS-bmFc%252Fhqdefault.jpg&w=1139&q=75)
Video tags
Subtitles
Russian
00:00:04
thanks everyone for joining today's00:00:05
webinar an introduction to l2l3 evpn00:00:08
services supported on the 7750 router00:00:11
my name is darren barron i'm the00:00:12
marketing and business support manager00:00:14
for learning services our presenter00:00:16
today is brian mckenzie subject matter00:00:18
expert from our learning services team00:00:21
as brian mentioned this webinar will be00:00:22
recorded and made available to you by00:00:25
email so please use the chat if you do00:00:27
have any questions and we'll do our best00:00:28
to get to them at the end00:00:31
today's webinar is actually based on00:00:33
some content from our newly released00:00:36
evpn services course00:00:39
op 0086 from the service routing00:00:41
certification program so i'm just going00:00:43
to give a quick overview of the program00:00:45
uh before we get start before brian gets00:00:48
into his material so this is our longest00:00:50
standing program we've had 30 000 plus00:00:52
learners 45 000 plus written exams at00:00:55
the last time we checked00:00:56
and over 12 30 12 000 certifications00:00:59
awarded so by far nokia's biggest00:01:01
certification program and probably most00:01:04
important00:01:05
um00:01:06
so from this we've got um over 1500:01:09
courses and workshops we've got00:01:12
three industry recognized certifications00:01:14
that i'll go quickly through in a minute00:01:16
these are developed by members of our00:01:18
team like brian and others working00:01:20
closely with internal nokia00:01:22
stakeholders to make sure we're using00:01:24
best practices and providing strong use00:01:27
cases and then obviously most of our00:01:29
courses are about 50 50 hands-on labs in00:01:32
theory and then we've got some different00:01:34
solutions00:01:36
for00:01:38
for different uh00:01:39
learning options00:01:42
so in terms of benefits from an00:01:43
organizational standpoint um i won't go00:01:45
into this in too much detail but00:01:47
there's a number of different reasons00:01:48
why training and certifying yourself and00:01:51
your organization00:01:53
can greatly improve um you know the00:01:55
speed in which you can release services00:01:58
your innovation00:02:00
um and just you know having a skilled00:02:02
and happy workforce and from an employee00:02:04
standpoint00:02:06
certainly a greater opportunity for00:02:07
promotion more confidence00:02:10
and then increased productivity so a00:02:12
number of different reasons why you00:02:13
might consider obviously keeping your00:02:15
skills up to date00:02:17
and and then validating them through00:02:19
certification00:02:20
and probably are one of our biggest uh00:02:23
customers based out of the us is a00:02:25
company called cox communications and i00:02:27
know they place a lot of emphasis uh on00:02:30
training and certification of their00:02:32
employees to ensure that they're00:02:33
delivering uh the best services they can00:02:38
um so the certifications in the program00:02:40
the nrs one is is your introductory00:02:42
certification um you know just learning00:02:45
the essentials of ip networking mvp vpn00:02:47
service routing00:02:52
oh look that we got somebody from cox00:02:53
here representing nice nice to hear from00:02:55
you good morning00:02:56
um and then uh moving on from that we've00:02:59
got our nrs2 certification which is00:03:01
obviously a level up and then after that00:03:03
we've got the service routing architect00:03:05
certification which which we've now00:03:07
included as part of that as an option to00:03:10
include evpn services and segment00:03:12
routing00:03:14
so here's just a quick look a diagram00:03:16
that highlights the certifications you00:03:18
can see here the nrs one00:03:20
basically uh you know one written exam00:03:23
and then you get into the nrs2 where00:03:25
you're choosing your routing protocols00:03:26
between isis and ospf00:03:29
and then bgp mpls and services00:03:31
architecture followed by a lab exam00:03:34
so this is where you get into a half day00:03:36
lab exam where you're doing00:03:37
configuration00:03:38
um00:03:40
and getting into a you know that's where00:03:42
it's really testing your knowledge and00:03:44
then on the service route routing00:03:46
architect it mainly builds on the uh00:03:48
nrs2 with the with a few differences00:03:50
where we've got a different uh a bgp00:03:52
course for internet routing and then we00:03:54
get into the vplan vprn and qos and then00:03:57
the choice of the recommended00:03:59
or a choice of electives00:04:01
so the courses or or the self-study are00:04:03
not mandatory you just have to do the00:04:05
exams but certainly your best bet is00:04:07
always going to be to00:04:09
you know prepare with the hands-on00:04:10
training00:04:14
and here's just a quick list i won't go00:04:15
through all the courses obviously but00:04:17
this is just a list of the current00:04:18
courses we have and we also have some00:04:20
workshops as well to help you prepare00:04:21
for the nrs2 lab exam00:04:25
so the different options we have right00:04:26
now00:04:27
from an instructor-led standpoint we've00:04:29
got our we've got an open public00:04:30
schedule which is uh00:04:32
which is most important for people that00:04:34
aren't saying you know you can't get00:04:35
away from the office or have six people00:04:37
attend a private delivery so you can00:04:40
attend a regularly scheduled public00:04:42
course which is available across00:04:43
multiple time zones in a virtual format00:04:47
we are getting back to some face-to-face00:04:48
but that's just reserved mainly for00:04:50
private or on-site deliveries00:04:52
and then we do sell these self-paced00:04:54
training so that the self the materials00:04:56
from the courses which are designed to00:04:58
be able to use self-paced as well to00:05:00
help you prepare for00:05:02
um written exams or just to self-learn00:05:04
and now we've we've recently made a00:05:06
numerous updates to our mysr lab00:05:08
environment so most of the courses from00:05:11
the nrs1 nrs2 and all but a few in the00:05:14
sra00:05:15
certification which we're just00:05:16
finalizing some updates on are all00:05:19
usable on mysr lab which means you can00:05:21
take the lab guide from the content and00:05:25
and complete the labs using mysrlab00:05:30
and then just tracking your tracking00:05:32
your success so you know when you're00:05:34
when you go to schedule your exams with00:05:36
nokia you didn't you do need a nokia00:05:37
account you don't have to be a customer00:05:39
or partner there is a guest access that00:05:41
you can use to get into00:05:43
uh to create a nokia account you can00:05:45
find that on our website when you go00:05:47
through the student portal00:05:50
just at the src site00:05:52
and and from there you can create your00:05:54
guest access and then00:05:55
automatically create a student portal00:05:57
account where you'll be able to track00:05:59
all your certification requirements00:06:00
schedule exams you can purchase course00:06:02
materials there00:06:04
find out what you need to do to renew00:06:06
your certifications00:06:07
and then00:06:08
and then from a written standpoint for00:06:10
the exams uh all exams are available00:06:13
um the written exams are available uh00:06:15
through pearson vue in person or or00:06:18
probably 90 percent of our registrations00:06:21
lately have been through onview which is00:06:23
the online learning00:06:24
and then the final piece for the nrs2 or00:06:27
the sra is the lab exam component which00:06:29
is also now both available uh virtually00:06:32
so those uh some locations are now00:06:35
available i00:06:36
actually i don't think we've opened any00:06:38
local sites yet for in-person exams but00:06:40
certainly they are available virtually00:06:46
and then if you do if you are coming00:06:48
from other background there is some00:06:49
opportunity to to get exemptions on the00:06:51
nrs2 for some of the uh i think isis00:06:53
ospf and bgp courses if you have some of00:06:56
these cisco or juniper certifications so00:06:58
it's worth checking out our exempt00:07:00
exemptions pages and then we have these00:07:02
uh composite exams that also allow you00:07:04
to combine exams00:07:06
um you know if you feel you've you've00:07:08
put in the work and you've and you've00:07:10
got the skills to go ahead and challenge00:07:12
several exams at once it's a quicker way00:07:14
to get through your certification00:07:18
and then we've got our practice exams on00:07:20
the lab exams we've got some what to00:07:22
expect documents and we also have some00:07:24
workshops available to help prepare for00:07:26
the nrs2 lab exams00:07:29
and then finally validating your00:07:30
accomplishments so we in the last year00:07:32
and a half i think now we've we've00:07:34
introduced badges through credly which00:07:36
is fairly common in the00:07:37
telecommunications industry so this is a00:07:39
great way to showcase your skills and00:07:41
also validate your skills because it is00:07:43
a verifiable and secure way00:07:46
you know to ensure that that00:07:48
you can improve your skills and it's not00:07:50
just written on a piece of paper00:07:51
somewhere00:07:53
using our student portal now you can00:07:54
access or print some nice new diplomas00:07:57
that we've created00:07:58
and then finally00:08:00
for anybody that's passed at least two00:08:02
exams in the src program you'll get this00:08:04
really nice plaque00:08:06
and then any00:08:08
exams after that that you pass you'll00:08:10
get00:08:11
tiles in the mail that you'll be able to00:08:13
stick right on there so i know brian's00:08:15
got a couple of them hanging up on his00:08:16
wall and they're00:08:18
they're quite nice00:08:23
and then just just the programs our team00:08:25
support so our team focuses on the you00:08:27
know as i said the service writing00:08:28
certification program00:08:29
the optical network certification00:08:31
program uh anybody who joined our00:08:33
webinar yesterday we talked about the00:08:35
data center fabric certification program00:08:36
which is a new program that's00:08:38
evolving quite fast00:08:40
and then00:08:41
our nsp program which is also going to00:08:44
evolve into a certification program00:08:45
later this year we're going to start00:08:47
introducing certifications and then from00:08:49
the sdn standpoint we've got the nuage00:08:51
networks virtuoso certification program00:08:55
and with that i'm going to hand it over00:08:56
to brian00:09:09
all right now i found my mouse i'm ready00:09:11
to go00:09:17
all right folks welcome uh00:09:19
so today what we're gonna go through is00:09:22
darren said is00:09:24
basically an introduction to our layer 200:09:26
layer 3 evpn services that we have00:09:30
now supported on the 7750 router00:09:33
this is a new course it is four days in00:09:35
length00:09:36
but00:09:37
we're just gonna hit the highlights of00:09:39
it at the end of it we'll have a00:09:41
demonstration i'm gonna create a couple00:09:44
different services one a layer two00:09:46
services that spans00:09:49
between two data centers and one only or00:09:51
three services that spans between two00:09:54
data centers00:09:55
and we'll00:09:57
talk about those00:09:59
when we get there maybe a little more00:10:01
detail on it00:10:04
so as you know00:10:06
brian sorry00:10:07
are you trying to share this slide00:10:09
am i not presenting uh while you're00:10:11
talking oh well00:10:13
that's bad then00:10:16
it's only bad if i let you go on for too00:10:17
much longer00:10:19
there we go00:10:24
thank you darren that was that's00:10:26
embarrassing00:10:28
well i know it's quite early for you so00:10:30
oh it's not not as bad as it's not as00:10:32
bad as yesterday and i didn't mess up00:10:34
yesterday not much00:10:35
[Laughter]00:10:38
uh00:10:40
all right so can everybody see my screen00:10:42
let's start00:10:46
okay so our vpn services00:10:50
as you know your service provider has00:10:52
a physical network00:10:54
and we utilize these vpn services to00:10:58
carve out00:10:59
portions of that physical network00:11:02
to00:11:03
satisfy a customer's requirement00:11:06
so these are just logical entities00:11:11
that exist on the physical network to00:11:14
support a customer's unique00:11:16
requirements for connectivity within our00:11:19
networks00:11:21
the big thing about the vpns is00:11:24
sorry to keep the customers traffic00:11:26
isolated from other customers00:11:31
so we have another number of different00:11:33
vpn services that we can support virtual00:11:36
private wire services which is a layer00:11:39
two but it's a point-to-point service it00:11:41
just has two endpoints between the00:11:43
customers sites00:11:45
we have our virtual private land service00:11:47
which is a layer two multi-point service00:11:49
so that we basically make the network00:11:52
look and behave like an ethernet switch00:11:56
and then we have our vprn or our virtual00:11:59
private routed network which is a layer00:12:01
3 multi-point services where again we're00:12:04
now making the network behave as if it's00:12:07
a unique router to instance00:12:10
when we create these different services00:12:13
they have to all be uniquely identified00:12:16
by a service id00:12:20
now in our standard services we have to00:12:22
have a couple different types of tunnels00:12:26
one is the transport tunnels that's the00:12:28
interconnection between the two rotors00:12:32
so each rotor has to have a transport00:12:34
tunnel that takes it through the core of00:12:36
the network and terminates on the far00:12:39
npe routers00:12:41
so wherever customers are coming into a00:12:43
service00:12:44
their traffic then has to go into this00:12:46
transport tunnel that will take to the00:12:48
far end where the other customer sites00:12:52
are located00:12:54
we have different types of transport00:12:56
tunnels one is mpls00:12:58
now the mpls labels can be distributed00:13:01
using rsvp ldp or segment routing00:13:05
another form of transport tunnel is our00:13:07
generic road encapsulation where we're00:13:09
just taking the information from the00:13:12
customer00:13:13
wrapping it up in an ip packet and00:13:16
sending it off to the far end00:13:19
and vxlan is another form of transport00:13:21
tunnel that we're going to see it in use00:13:23
today00:13:25
vxlan again we're wrapping it up into an00:13:28
ip packet but it also in the vxlan00:13:31
header is a vax lan network identifier00:13:34
that identifies which service00:13:37
that this traffic belongs to00:13:41
so depending upon the type of transport00:13:44
tunnel we're using we're going to have00:13:47
for example an mpls tunnel00:13:50
will have one mpls transport tunnel that00:13:53
can carry multiple services to the same00:13:56
destination00:14:00
so to keep that cup traffic isolated in00:14:03
that specific transport tunnel we would00:14:06
create a second tunnel00:14:07
that resides in that transport tunnel00:14:10
and these are referred to as our service00:14:12
tunnels00:14:14
the labels that are utilized for the00:14:17
service tunnels are negotiated directly00:14:19
between the two pes and they're00:14:22
utilizing either targeted ldp or00:14:25
multi-protocol bgp to exchange these00:14:28
labels00:14:30
so in our00:14:32
standard deployment00:14:34
of a service00:14:36
we have multiple control plane protocols00:14:39
we need a routing protocol that tells me00:14:41
how to get from p e pe1 to pe200:14:45
we need a signaling protocol that will00:14:47
set up my mpls transport tunnel now we00:14:50
need a second signaling protocol that00:14:52
will set up my service tunnels00:14:56
so00:14:57
we need to be able to reduce00:15:00
this type of signaling and that's what00:15:02
evpn does for us00:15:05
evpn gives us this unified topology or00:15:09
umbrella that encompasses basically all00:15:12
the services that i can offer00:15:15
so we have our layer two services and00:15:18
e-line or ethernet line which is our00:15:20
virtual private wire service00:15:23
we have our e-land or ethernet lan which00:15:25
is our vpls service00:15:28
and we offer also the e-tree which is00:15:31
basically if you think about it as a hub00:15:33
and spoke type service where the root00:15:36
can talk to all the branches00:15:39
but if the branches want to communicate00:15:41
to each other they must go through the00:15:43
root to get there00:15:46
in our layer 3 services we have our00:15:49
standard vprn00:15:51
but we also can support our next00:15:53
generation multicast vpn00:15:57
so all of these services can fall under00:16:00
the umbrella of evpn00:16:05
so to make life easy for us00:16:09
we use multi-protocol bgp00:16:13
to exchange the network information00:16:16
required for these services00:16:19
there are a number of different route00:16:21
types to find depending upon the type of00:16:24
service that you're configuring00:16:27
we exchange the evpn routes between all00:16:30
the pes that are configured as my peers00:16:33
in my multi-protocol bgp configurations00:16:40
so the route types that we used as i00:16:42
said depend upon the services that00:16:44
you're going to see00:16:47
so here are00:16:48
five of the different route types i'm00:16:50
only showing up the ones that we use for00:16:53
a unicast scenario00:16:55
when we get into the next generation00:16:57
multicast scenario there's another four00:17:00
or five route types associated to that00:17:03
so we're only looking at the ones here00:17:06
for our unicast type scenario00:17:09
so our first one is the ethernet auto00:17:11
discovery so it's a layer it's used in a00:17:14
layer two service that supports00:17:16
multi-homing00:17:19
what we do is we identify00:17:22
to the far npes all the remote pes that00:17:26
this specific local router is connected00:17:30
to an ethernet segment that is00:17:32
multi-homed00:17:35
then we have our mac ip advertisement00:17:38
this is a step up from the standard vpls00:17:41
service00:17:42
a standard vpls service builds its00:17:45
forwarding database based upon the data00:17:48
plane it has to receive an ethernet00:17:51
frame to be able to put that source mac00:17:54
address into its forwarding table00:17:57
with an evpn as soon as a peu learns a00:18:02
local mac address00:18:04
they can generate a mac advertisement00:18:06
and send that thing send that mac00:18:09
address to all the pe's participating in00:18:11
the service00:18:13
so that those remote pes can now build00:18:15
their forwarding table without waiting00:18:18
for a data pack00:18:22
we have our inclusive multicast ethernet00:18:25
tag or our imet00:18:27
this is utilized to discover all the pes00:18:30
that are participating in a specific00:18:32
service00:18:33
that's its first function00:18:37
so as soon as i create a service on a pe00:18:40
router and that service becomes00:18:42
operational00:18:44
multi-protocol bgp will send this type 300:18:47
route to all the pe's identifying the00:18:50
service that we're utilizing00:18:52
and the label you need to use to send00:18:55
traffic to me and that's for my00:18:58
broadcast unknown destination and00:19:00
multicast traffic00:19:02
so all the remote pe's can build their00:19:05
40 or their flooding trees00:19:08
based upon these imet tanks00:19:14
the type 4 route or my ethernet segment00:19:17
wrote again is used in a layer 2 service00:19:19
in the multi-homing scenario00:19:21
this is exchanged between the pes that00:19:25
are all connected to the same ce device00:19:28
using the same ethernet segment00:19:32
we're going to00:19:34
use this to discover all of the pe's00:19:36
that are connected to this ethernet00:19:38
segment00:19:39
and identify which pe will be the00:19:42
designated forwarder for this ethernet00:19:45
segment and we'll see the various roles00:19:48
dependent upon the type of redundancy00:19:50
that's used in this ethernet segment00:19:54
if it's a00:19:55
active standby scenario or it's an all00:19:58
active scenario we'll see that in a bit00:20:03
the one route type for my layer 3 vpn is00:20:06
the ip prefix00:20:07
and as it says here that's its job is to00:20:10
generate an update with all the ip00:20:14
prefixes that are locally connected to00:20:16
this vprn service00:20:26
my data plane00:20:29
so we can00:20:30
evpn allows us to use00:20:33
multiple technologies in the core of the00:20:35
network00:20:37
so the ingress pe's job is to00:20:40
encapsulate the customer's traffic00:20:43
with a label of some kind that uniquely00:20:45
identifies each service so that would be00:20:48
referred to as our service label00:20:51
however in an evpn environment that00:20:53
service label could be an mpls label00:20:57
or it could be a vxlan network00:20:59
identifier00:21:01
so evpn gives us the flexibility to use00:21:04
both mpls or vxlan00:21:08
so we don't have to00:21:10
modify our network00:21:13
we're going to encaps the data00:21:15
encapsulate the data and then tunnel it00:21:17
between the pe so00:21:19
we first put the label on it that00:21:21
identifies the specific service then we00:21:24
take that00:21:25
labeled packet and put it in the tunnel00:21:28
that takes me from my local p e to the00:21:30
remote p e00:21:33
and so we can do evpn over mpls00:21:37
we can do evpn over vxlan00:21:48
so let's talk first about our layer two00:21:50
service our elan service00:21:57
some new terminology for some old00:22:01
systems00:22:03
so first thing is the broadcast domain00:22:06
since it's a layer two service all the00:22:08
connections are in a single broadcast00:22:11
domain00:22:13
these broadcast domains are all00:22:15
associated to an evpn instance00:22:18
identifier00:22:20
so for those who have been playing with00:22:22
vpls a long time00:22:25
think of the evpn instance identifier as00:22:27
the service id number00:22:31
the mac vrf is the forwarding table that00:22:34
we're going to use inside each one of00:22:37
these instances to forward traffic in00:22:39
there so it's going to contain all the00:22:42
mac addresses of the ce's00:22:45
that are connected into this specific00:22:47
service00:22:53
so when you first turn up a service and00:22:56
it becomes operational the first route00:22:59
that it's going to advertise is that00:23:01
type 3 imet route00:23:03
it's going to identify00:23:06
the service that you're associated to by00:23:08
using the route target that's associated00:23:11
to that service00:23:12
and the label that you can use00:23:15
to access this service for your00:23:17
broadcast unknown and multicast traffic00:23:22
so each one of the pes will generate00:23:25
this imet route so each pe can now00:23:29
create what's called referred to as the00:23:31
flooding list00:23:32
if i get a broadcast message00:23:35
coming in from ce1 to pe1 how do i get00:23:39
it to the vpls service on pe2 and pe300:23:43
well i'll replicate it00:23:45
afix the appropriate label and then send00:23:48
it out to pe200:23:50
mpe 3 through the transport tunnels00:23:53
whether that transport tunnel be mpls or00:23:56
vxlan00:23:58
as i said this label00:24:01
could be an mpls label it could be a vni00:24:05
pxlan network identifier00:24:13
then the next thing is00:24:15
learning a local mac address so00:24:18
let's assume we have a protocol like00:24:21
lldp running between the pe and the ce00:24:24
device so i'm going to learn them ce's00:24:27
mac address00:24:29
as soon as i learn that ce mac address00:24:31
i'm going to generate a type 2 wrote a00:24:34
mac ip advertisement now i'm going to00:24:36
advertise this mac address to the pe00:24:39
devices00:24:41
and they can now start building their00:24:43
forwarding table without actually having00:24:45
to receive data00:24:48
from ce100:24:50
and each pe will do the same as they00:24:53
learn a local mac address they will00:24:55
generate a type 2 route send it out00:25:03
the other thing we see here it says mac00:25:05
ip advertisement00:25:08
we can turn on proxy onto these vpls00:25:12
services00:25:14
and then when i advertise out the mac00:25:16
address for ce1 i'll also advertise out00:25:20
its ip address00:25:22
and then the remote pe's can build their00:25:24
proxy arp tables00:25:27
so if ce2 for example creates an arp00:25:31
message for the mac address for ce100:25:35
he'll generate that up to pe2 pe2 now00:25:39
has the mac ip pairing of ce100:25:42
so pe2 can respond as a proxy for ce100:25:47
and give ce2 the mac address00:25:50
that it will allow it to communicate to00:25:52
ce100:25:56
thereby reducing the amount of broadcast00:25:59
traffic that goes through the courier00:26:01
network00:26:05
we talked about our layer two services00:26:07
whether they be e-line or e-land support00:26:11
multi-homing00:26:13
we have two options with the00:26:15
multi-homing we have a single active00:26:18
where one pe is in charge of forwarding00:26:21
all traffic to and from the ce device00:26:24
and then we have the all active mode00:26:27
where all the pe's that are connected to00:26:29
this ethernet segment00:26:31
will be able to forward traffic to and00:26:33
from the device00:26:35
okay00:26:40
now to try to control this00:26:42
we have our ethernet segment concept and00:26:45
the ethernet segment is all the links00:26:47
that connect00:26:49
one or more pe's to a ce device00:26:55
so we have to uniquely identify this00:26:58
ethernet segment so that's how we have00:27:00
our ethernet segment identifier that00:27:02
again00:27:03
gives us a common00:27:05
number that we're going to use00:27:07
on both pes here that are connected to00:27:10
it00:27:11
now depending upon00:27:14
the redundancy method if we go back00:27:17
see if we have an all active then we're00:27:19
going to need the ce to be configured00:27:22
with a leg so that we prevent loops from00:27:25
forming00:27:29
so we have to identify whether or not00:27:31
we're doing an all active scenario and00:27:33
then if we are doing an all-active then00:27:35
that ce has to have a lag configured on00:27:38
these links that connect up to the pe00:27:40
devices00:27:46
so for multi-homing we talked about them00:27:48
earlier we have two different types of00:27:50
right the auto discovery routes and the00:27:53
ethernet segment routes00:27:55
and then the auto discovery routes are00:27:59
have two subtypes associated to it one00:28:01
is the auto discovery per ethernet00:28:04
segment and then the auto discovery per00:28:06
evpn00:28:09
so let's take a look at these00:28:13
so the first one we'll look at is the00:28:15
auto discovery per ethernet segment00:28:17
route00:28:19
so this is identified or sent out by all00:28:21
the pe's that are connected to00:28:24
an ethernet segment00:28:26
in there they're identifying the00:28:28
redundancy mode for that ethernet00:28:30
segment whether it's all active or00:28:32
single active00:28:34
and it also includes an ethernet segment00:28:36
identifier label00:28:39
this is for split horizon between the00:28:41
pe's that are connected to that ethernet00:28:45
segment00:28:49
excuse me00:28:53
we'll see that later on just to make00:28:55
sure that we don't have a loop00:28:57
we'll see this in a bit but that's why00:29:00
we send out this label00:29:04
now an auto discovery per evpn instance00:29:08
or so we have the evidence00:29:12
we're going to advertise an evpn00:29:14
instance that is associated to this00:29:17
ethernet segment so here is my service00:29:20
id and my service id is connected to00:29:22
this ethernet segment00:29:25
now we're also identifying00:29:28
what is my role00:29:30
for that ethernet segment so those pe's00:29:33
that are connected to that ethernet00:29:35
segment will identify their role00:29:37
am i a primary or am i a standby00:29:41
depending upon the redundancy00:29:43
if i have multiple pes connected into an00:29:46
all active scenario then they identify00:29:49
to all the remote pes that they're all00:29:51
primary00:29:53
if we're in a single active then i have00:29:56
a primary and a backup00:29:59
so when they're all active the remote00:30:01
pe's can create what's called an00:30:04
aliasing list00:30:05
it's just identifying that it can send00:30:08
traffic to one if not all of the pes00:30:11
connected to the ethernet segment00:30:14
so if that vp vpn is configured for00:30:18
load balancing or ecmp00:30:20
it can now load balance between all the00:30:23
pe's that are connected to that ethernet00:30:25
segment00:30:28
when i have a single active es one is00:30:30
identified as the primary he is the00:30:33
designated forwarder for that ethernet00:30:35
segment00:30:36
and all the other pes00:30:39
are not designated forwarders so they00:30:40
become the backup00:30:43
so they all the other pe s go into the00:30:45
remote pe's backup list so should the00:30:48
primary fail we can00:30:51
then ident we can identify a backup to00:30:54
it00:31:00
the type 4 ethernet segment wrote00:31:03
is sent between the pes that are00:31:05
connected to the ethernet segment00:31:09
we ensure that it's only used by the00:31:12
pe's that are connected to the ethernet00:31:14
segment by00:31:16
automatically deriving the route target00:31:19
from the ethernet segment identifier00:31:22
so pes that are not connected to this00:31:25
ethernet segment00:31:26
will not import the routes00:31:29
only the pe's that are connected to this00:31:32
ethernet segment with that ethernet00:31:34
segment identifier will import the roads00:31:38
we use this00:31:39
ethernet segment wrote00:31:42
to00:31:43
identify who will be the designated00:31:46
forwarder00:31:51
so we saw that we can have a designated00:31:53
we're going to have a designated00:31:55
forwarder for the ethernet segment00:31:57
whether00:31:58
it's all active or whether it's a00:32:00
standby00:32:02
active scenario we're still going to00:32:04
have a designated forwarder so depending00:32:07
upon the redundancy role00:32:10
we have different functions that the00:32:12
designated forwarder and non-designated00:32:14
forwarder will use00:32:18
so in an all active scenario00:32:20
my designated forwarder can forward all00:32:22
the traffic and receive all the traffic00:32:25
whether it's from the ce or from00:32:28
the remote pe's00:32:31
in the non-designated forwarder00:32:34
he can send and receive unicast traffic00:32:38
to and from the ce00:32:41
he can receive multicast traffic from00:32:44
the ce00:32:45
but he cannot transmit multicast traffic00:32:49
to the ce this is one of the ways we00:32:52
prevent loops00:32:54
only the designated forwarder can treat00:32:57
send multicast broadcasts or unknown00:32:59
destination traffic to the ce00:33:02
just excuse me for a second00:33:16
sorry i had a tickle in my throat had to00:33:18
get rid of00:33:21
in a single active scenario00:33:28
as you can see the non-df just goes into00:33:30
standby he cannot receive transmit00:33:34
any kind of traffic00:33:36
only the df or the designated forwarder00:33:39
can send and receive traffic00:33:48
so let's take a look at an all active00:33:50
multi-homing scenario and this aliasing00:33:54
as seen in all active environment00:33:57
we have first00:33:59
advertisement about the ethernet segment00:34:02
and the redundancy model of that00:34:04
ethernet segment so we are advertising00:34:06
that to all the pe's00:34:11
then we're advertising the evi00:34:14
okay my service id number00:34:18
that is connected to that ethernet00:34:20
segment and then how to reach me00:34:24
if you want to use me as access to the00:34:27
ce00:34:29
or this service that's connected to this00:34:32
ethernet segment00:34:34
so on my remote pe00:34:37
we now know that00:34:38
there is an e ethernet segment out there00:34:41
we've learned that through the auto00:34:43
discovery per ethernet segment row00:34:46
and we know to reach this ethernet00:34:48
segment i can send it to either pe300:34:52
or pe1 using these labels00:34:58
now00:34:59
ce00:35:01
he has a lag configured he sends a00:35:03
packet up into pe one of some kind00:35:06
whether it's controller data don't care00:35:09
pe one is going to advertise that00:35:11
there's a mac00:35:13
and this is the mac address00:35:15
and it's associated to ethernet segment00:35:17
identifier 1.00:35:19
we put that into the tape forwarding00:35:22
table of pe 2 saying to reach this mac00:35:24
address i got to get to esi100:35:28
to get to esi1 i can send it either to00:35:31
label or either to pe1 or pe300:35:44
we talked about the split horizon00:35:46
earlier and this is the auto discovery00:35:48
per ethernet segment wrote00:35:51
so as these devices were connected to00:35:55
the segment and a vpls service came up00:35:57
on there we identified that we are00:36:00
connected00:36:01
to this service and we sent that auto00:36:03
discovery per ethernet segment route to00:36:06
all the pe's00:36:08
well when i sent it from pe 3 to pe 1 i00:36:11
identified a label00:36:13
that he should use00:36:16
when he wants to send me any broadcast00:36:18
or unknown multicast traffic00:36:21
that is associated to this ethernet00:36:23
segment00:36:24
that way when i receive this esi label00:36:28
as part of the packet00:36:30
i know not to forward it out the leg00:36:33
i would just get rid of it00:36:37
because that's a label that i sent this00:36:39
is how we're achieving split horizon00:36:42
so when the00:36:44
pe in this case we we have the all00:36:47
active scenario00:36:49
the ce is sending out a broadcast frame00:36:52
the service has to set that frame into00:36:55
his flooding list now the flooding list00:36:58
includes all the pe's that are00:37:00
participating in the service00:37:03
i know pe 2 is not connected to the00:37:07
ethernet segment because he did not send00:37:09
me an ethernet segment identifier label00:37:11
so i'll just add on the label that00:37:13
identifies his service and sent him the00:37:16
traffic00:37:19
however pe300:37:21
did identify that he's connected to the00:37:23
same ethernet segment i am and this is00:37:26
my esi label that he sent me00:37:29
so i will wrap the traffic up in the esi00:37:32
label and then put the flooding list00:37:34
label that he sent me on the imet00:37:37
and send that off to pe300:37:40
pe3 will pop this label00:37:44
interrogate the esi label see that it's00:37:47
associated to this ethernet segment so00:37:50
he will not forward the traffic out this00:37:53
ethernet segment00:37:56
the reason why we send00:37:58
pe 3 that label00:38:01
is because that vpls could have another00:38:04
ce hanging off of it right here00:38:07
so we have to send him the flooding00:38:11
traffic00:38:12
we just have to identify00:38:16
don't send it out the leg if you have00:38:18
another ce you can send it to him don't00:38:21
send it out the leg00:38:22
so that's why we have the esi label00:38:28
you won't send it out the leg but he00:38:29
will send it to this other ce that's00:38:32
part of that vpls service00:38:45
so the df as we talked about the des00:38:48
native forwarder00:38:50
is elected using that uh ethernet00:38:53
segment route00:38:55
in a single active multi-home in the df00:38:58
rule00:38:59
is everything all traffic must go00:39:01
through the designated forwarder00:39:04
the sat the non-designated forwarder00:39:06
brings their saps00:39:08
that service access point that connects00:39:10
to the ce they bring that to00:39:13
operationally down00:39:15
so any traffic that's going to or from00:39:18
the ce00:39:20
has to go through the designated00:39:22
forwarder00:39:29
so our single multi-homing advertised00:39:31
routes00:39:32
again we both send out the ad per00:39:35
ethernet segment identifying that we are00:39:38
both connected to the ethernet00:39:40
segment and that it is in single active00:39:45
and then our role for that ethernet00:39:47
segment00:39:48
and then we will send out the ad per evi00:39:54
okay so pe2 will receive all this00:39:56
information you'll see that the role of00:39:59
pe1 is the primary the role of pe 3 is00:40:03
the backup00:40:04
so whenever he receives a route00:40:07
a macro00:40:09
we're going to see that in his00:40:11
forwarding table00:40:13
i received the mac address associated to00:40:16
esi100:40:18
to get to esi 1 i must send it to pe100:40:22
using this label he is the primary00:40:25
pe 3 is a backup should the primary fail00:40:30
then i can send the traffic to00:40:32
pe300:40:37
now pe3 since it's the non-designated00:40:40
forwarder he's learned this mac address00:40:43
as well because when we sent it out00:40:46
okay i sent out this mac address i sent00:40:49
it to pe2 but i also sent it to pe 300:40:52
because he is a bgp peer00:40:56
so pe3 puts that mac information into00:41:00
his forwarding table it's mapping to00:41:02
esi1 which is ethernet segment but since00:41:06
he's the non-df and the samp is00:41:08
operationally down his mapping of esi00:41:11
one is to pe100:41:14
so if he gets traffic destin00:41:18
example again we'll go back to my pen00:41:23
and we'll draw my other ce here00:41:27
and connect it see how fancy drawing i00:41:30
am00:41:31
good look at that00:41:32
we'll connect it into the vpls so if00:41:35
this ce wants to send traffic00:41:38
to ce1 he sends it to the v into the00:41:41
vpls service the vpls service sees that00:41:45
ce1's mac address is associated to the00:41:48
esi00:41:50
pe3 actually has to send the traffic up00:41:53
to00:41:54
pe one00:41:56
so that pe one can forward the traffic00:41:59
to the ce00:42:14
so vpls cross ip mpls networks00:42:19
but we can have an interconnection00:42:21
between data centers00:42:26
so this data center could be vxlan00:42:30
this data center could be vxlan00:42:33
two separate data centers00:42:35
we're interconnecting them across an00:42:37
mpls core00:42:39
now this mpls core00:42:41
could be evpn or non-evpn00:42:45
we can still interconnect them across00:42:47
the course00:42:49
so we can allow vxlan00:42:52
vxlan and then interconnect them across00:42:55
an ip mpls core whether it be00:42:59
evpn00:43:01
or non-evpn in this core network00:43:07
and we can do that for all layer 200:43:10
services00:43:17
so our layer 3 service00:43:21
bprn service00:43:23
we've got a couple different 40 models00:43:26
that evpn has defined00:43:29
for our vprn services there's the00:43:32
asymmetric and symmetric 40 models and00:43:35
we'll just briefly touch on them all00:43:40
so the asymmetric forwarding model00:43:43
means that all00:43:45
forwarding layer 3 forwarding decisions00:43:47
are local00:43:49
so as you can see here i've got host 200:43:52
that's connected to the 1.0 24 network00:43:56
and host 4 that's connected to the 4.000:43:59
24 network00:44:01
with this00:44:03
on pe 1 i must have a routed instance00:44:06
for the vpls00:44:09
that connects to pe1 that belongs to00:44:11
that four network00:44:14
doesn't seem too bad but let's bring in00:44:17
my trusty pen again00:44:21
let's connect a third00:44:23
bpls service00:44:26
connecting to another ce device00:44:29
for another network00:44:32
well that means that over here00:44:34
we must have another vpls service for00:44:38
that subnet00:44:39
and we have another tunnel coming across00:44:43
because everything is forward through00:44:45
the network via layer 2. so every time i00:44:48
add in another00:44:50
subnet00:44:52
every pe must be configured with a vpls00:44:56
service that belongs to that subnet00:44:58
even though they don't have00:45:01
a host associated to that subnet00:45:06
so as you can see over here vprn 10 only00:45:09
has a single subnet connected to it as00:45:12
far as a host00:45:13
but it has two other subnets that it has00:45:16
to have configured on it to get to the00:45:18
vpls service that will allow us it to00:45:21
communicate to the local host00:45:24
this is the asymmetric forwarding model00:45:26
so it's not a scalable solution00:45:36
the other one is the symmetric 40 model00:45:40
this is giving us00:45:42
uh vrf to vrf prefix runs and we're00:45:45
going to be using that whether we're00:45:47
going interfaceless00:45:49
interface full and when we get into the00:45:51
interface full we have numbered or00:45:53
unnumbered we're going to be using these00:45:56
ip prefix routes00:46:00
we're going to talk touch on these00:46:02
symmetric models and we'll touch on all00:46:04
three of these the interfaceless00:46:06
interface full numbered and interfaceful00:46:08
unnumbered and we're just going to touch00:46:10
on them in the high level00:46:12
the interface list model00:46:15
is something that we're all familiar00:46:16
with this is basically how our standard00:46:18
vprn works00:46:21
we have a local subnet we generate an ip00:46:24
prefix route and we advertise it to the00:46:27
vprn on the far mpe he populates that00:46:31
into his routing table00:46:32
so we don't require any mac ip routes to00:46:35
be exchanged alls we need is an ip00:46:38
prefix route and we can exchange the00:46:41
information across00:46:43
that's our interface less vrf to vrf00:46:46
model00:46:48
but let's have some flexibility00:46:52
let's go to the interface full numbered00:46:55
vrf to vrf model00:46:58
what we're going to do is now create00:47:01
what's referred to as a supplementary00:47:03
broadcast domain some of you may know00:47:05
this as a backhaul00:47:08
vpls service00:47:10
and it is a layer 2 service00:47:13
it's different from00:47:16
the asymmetrical side because now00:47:20
if i bring in00:47:23
that other vpls service00:47:28
with my other ce so i bring in a second00:47:31
or subnet here on vprn 10.00:47:38
laser pointer thank you00:47:40
i don't have to do any configurations in00:47:42
here my supplementary broadcast domain00:47:45
will still be used to interconnect the00:47:47
vprns and i'll still be able to send00:47:49
stuff across00:47:51
if i bring in a third00:47:54
router00:47:56
third pe device00:47:58
all i have to do is interconnect them00:48:02
using the supplementary broadcast domain00:48:05
so he can bring in as many subnets as we00:48:07
want from this other pe00:48:11
and we don't have to do anything like we00:48:13
had to do in the asymmetric version00:48:17
we only have to maintain the00:48:20
supplementary broadcast domain00:48:24
and any routed vpls or host00:48:26
connectivities that we want on our pe00:48:32
okay so my supplementary broadcast00:48:34
domain has no saps what it has is an uh00:48:37
integrated routing and bridging00:48:38
interface that00:48:40
logically connects the vprn to the00:48:43
supplementary broadcast domain just as00:48:45
it does logically connecting the eroded00:48:47
vpls to the vpn00:49:01
so in my interface full number two00:49:05
model of our routes that we're00:49:06
advertising we're advertising two00:49:08
different routes00:49:10
so we'll take a look from pe two's00:49:12
perspective he has the 10 10 ford00:49:15
network connected to him00:49:17
so he's going to generate two roads the00:49:19
first route is my ip prefix route00:49:22
and it's identifying the gateway ip00:49:25
address00:49:26
as the interface that connects it the00:49:29
vprn to the supplementary broadcast00:49:31
domain00:49:33
and he's going to advertise that over to00:49:35
the vprn 10.00:49:39
now there's a mac address associated to00:49:41
this interface and that interface can00:49:43
that mac address and interface combo00:49:45
belongs to the supplementary broadcast00:49:48
domain so it will advertise a mac ip00:49:51
wrote00:49:53
saying if you wanted to get to 10 10 1000:49:56
100 3.2 this is the mac address that you00:49:59
have to use and this is the vni00:50:02
for this00:50:03
uh supplementary broadcast domain or an00:50:06
mpls label for the supplementary00:50:08
broadcast domain depending upon what is00:50:11
the core i'm using00:50:13
and so my vprn gets the route00:50:17
and identifies that he has to get to00:50:19
10-103200:50:21
which is local here00:50:25
my supplementary broadcast domain gets00:50:27
the mac address associated with 10 103.200:50:31
so when the vprn goes to send it he00:50:34
knows to get to this address he must00:50:36
wrap it up into this mac ip or this mac00:50:39
address00:50:41
and that mac address is forwarded by the00:50:44
supplementary broadcast domain to this00:50:46
interface00:50:48
so i'm still doing local routing00:50:51
still doing layer 2 across the core00:50:55
but i don't have to have multiple00:50:57
instances00:50:58
of a vpls00:51:00
that i know don't have a local host for00:51:03
i have just the one supplementary00:51:06
broadcast domain that can interconnect00:51:11
so the data flow00:51:14
we have our payload00:51:16
we have our source mac address which is00:51:18
the mac address of vpr and tens00:51:21
interface00:51:22
we have the destination mac address00:51:25
which is the mac address of this bprn00:51:28
interface00:51:30
we have the vni00:51:32
that identifies00:51:34
the service for my supplementary00:51:36
broadcast domain00:51:38
and then the source ip address be the00:51:42
system interface of pe100:51:44
destination ip address be the system00:51:47
interface of pe200:51:50
and we send our traffic across my vxlan00:51:53
core00:52:02
now if you get into a data center and00:52:04
you take a look at a data center00:52:06
architecture there is a ton of00:52:08
point-to-point interfaces00:52:10
and a lot of these interfaces you have00:52:12
to put numbers on them ip addresses on00:52:15
them the same thing with my vprn00:52:18
instances we could have multiple vprn00:52:21
instances and i would have to consume ip00:52:24
addresses00:52:26
on all the interfaces that are00:52:28
interconnected to my sbd00:52:31
so we have to take a look at this00:52:33
information and say is it worth it for00:52:36
me to put an ip address on an interface00:52:41
that's never going to get advertised00:52:43
outside of my network00:52:45
and the answer is no i don't have to00:52:47
consume ip addresses so we have the00:52:50
interface full unnumbered vrf to vrf00:52:53
models00:52:56
configures up basically the same00:52:59
the only difference is i don't assign an00:53:02
ip address00:53:04
to these interfaces that connect the00:53:06
vprns to my supplementary broadcast00:53:09
domain00:53:10
therefore i'm not consuming as many ip00:53:13
addresses as i would in a normal00:53:15
situation00:53:19
so when i advertise out my prefix route00:53:22
we go back to pe number two again and i00:53:25
advertise up my prefix00:53:28
i advertise out the mac address that's00:53:30
associated to that prefix and the spd00:53:33
and the vni that's associated to my spd00:53:37
the gateway ip address is zero00:53:41
i also advertise the mac out to mac00:53:44
address on a mac ip route to the sbd00:53:47
itself because remember this prefix00:53:49
route is only for the vprn 10.00:53:52
so i have to send out this macro to sbd00:53:57
it has a mac address in it there's no ip00:54:00
associated to it but i do have the mpls00:54:03
or vni label to reach the far end00:54:08
supplementary broadcast domain vpls00:54:11
service00:54:15
so00:54:16
now when vprn 10 wants to send traffic00:54:20
from host number two to host number four00:54:23
the actual00:54:25
data packet encapsulation00:54:27
is basically identical i have my payload00:54:31
i know my destination mac address is mac00:54:34
address 2 my source mac address is mac00:54:37
address one00:54:39
i have the vxlan network identifier that00:54:42
identifies this sbd on the far end pe00:54:47
i have my source ip address is still my00:54:49
system interface my destination ip00:54:52
address is still the system interface of00:54:54
pe200:54:57
so from a packet forwarding standpoint00:55:00
going to an interface full unnumbered00:55:02
model00:55:04
remains the same00:55:07
we are just saving the ip address space00:55:11
that we would have used to interconnect00:55:13
the vprns to the supplementary broadcast00:55:16
to me00:55:23
so00:55:24
we take a look at the different00:55:26
interface list models that we have and00:55:29
the interface full models that we have00:55:30
in our symmetric routing the interface00:55:33
list does it reduce the number of evpn00:55:35
routes00:55:36
yes it does00:55:38
because alls i send is the ip prefix00:55:41
back and forth between the vprns00:55:46
ip addresses configured on the core irb00:55:50
well there are no ip addresses00:55:52
configured on the core irb because there00:55:54
is no core irp same with the massive00:55:57
withdrawal to to a failure in my00:55:59
recursive lookups00:56:02
but it is00:56:03
has to be supported it's standard00:56:06
or if you're running the interface list00:56:08
it is a mandatory that you support this00:56:11
interface full numbered reduce the00:56:14
number of evpn routes absolutely not00:56:18
i'm sending two routes every time00:56:22
i have to have a since it's numbered i00:56:24
have to have an ip address on my core00:56:27
irb00:56:28
but we do have the mass withdrawal00:56:30
because of the recursive failure00:56:34
it is mandatory if you're running00:56:36
interface full to support this numbered00:56:39
setup00:56:41
if you're running interface full00:56:43
unnumbered00:56:45
okay we're still sending out the two00:56:47
routes so no we do not reduce the evpn00:56:50
routes but now we're not consuming any00:56:52
ip addresses we still support the masses00:56:55
withdrawal00:56:56
but the interface full unnumbered is an00:56:59
optional support for interface full00:57:02
symmetrical routing00:57:13
now my evpn00:57:15
virtual private router networks across a00:57:18
core again i can have00:57:20
a vxlan data center00:57:23
another vxlan data center00:57:26
and a core00:57:27
mpls network00:57:29
we can connect these across that core00:57:32
mpls cook network whether it's an evpn00:57:36
mpls00:57:37
or a non-evpn mpls as we're showing here00:57:41
by using the ipvpn routes00:57:52
so we got a couple demonstrations coming00:57:55
up where i'm going to go in and00:57:56
configure some equipment the first one00:57:59
i'm going to be taking a layer 2 service00:58:02
across an evpn mpls core00:58:06
so in the interest of saving time00:58:09
this side of the network's already been00:58:11
configured i'm just going to configure00:58:13
up pe1 and pe3 and connect it across00:58:17
and we'll see that these two devices can00:58:20
communicate together00:58:22
once i've done that00:58:24
then i'm going to slip into another lab00:58:27
and we're going to configure a layer 300:58:29
evpn service across an mpls core that is00:58:33
not evpn capable it's an ipvpn core00:58:38
and then say again in the interest of00:58:40
saving time this side of the network's00:58:42
already been pre-configured and so has00:58:45
pe1 so i'll only be configuring up pe 300:58:49
here00:58:51
but before i get into that00:58:53
i would say you've been sitting long00:58:55
enough how about we take a little break00:59:00
say about 10 minute break00:59:03
and then we'll come back and i'll be set00:59:05
up ready to go00:59:07
brian did you want to00:59:08
i mean after the break is fine look at a00:59:10
couple of these questions as well i00:59:12
think there's and obviously i can't00:59:13
answer any of them so00:59:15
all right so yeah we can do the00:59:17
questions yeah just a couple yeah just00:59:19
so we don't get too bombarded at the end00:59:21
there's all right yeah we can do that so00:59:23
let's take 10 minutes come back we'll00:59:24
hit the question list00:59:26
and then i'll take you through a quick00:59:29
run through00:59:30
player two and then a layer three00:59:32
service00:59:33
awesome thanks brian00:59:37
hey everyone you may have noticed i put00:59:39
the link from yesterday's webinar in the00:59:41
uh00:59:43
in the chat00:59:44
so um i'm also gonna i'll also send an00:59:46
email i'm gonna try and get out uh today00:59:48
or tomorrow00:59:50
uh no today tomorrow's canada day won't00:59:52
be in tomorrow00:59:53
[Laughter]00:59:55
well being a canadian in the us i get00:59:58
friday and monday off01:00:00
sweet01:00:04
how do i get that i try to tell him in01:00:07
that all the time being a canadian in01:00:09
the u.s i should be able to get both01:00:11
homes oh that doesn't actually work01:00:13
it doesn't work that way but i keep01:00:15
trying01:00:21
all right01:00:23
out there what do we have for questions01:00:24
my friend oh right yeah it might be01:00:27
better if you take a quick look through01:00:29
it just because um01:00:31
there's kind of a chain going on01:00:35
let's see01:00:37
oh yeah there's a lot01:00:38
whoa01:00:39
yeah so i think it starts with what kind01:00:41
of multicast01:00:43
tree for type01:00:44
n3nl nlri is supported01:00:48
ingress replication mldp01:00:52
okay yeah exactly uh by default it is01:00:57
ingress replication01:01:00
but you can configure it for01:01:02
multi01:01:03
mldp or01:01:05
point to multipoint rsvp if you wish to01:01:08
do it by default it is ingress01:01:10
replication01:01:14
okay sorry i'm just trying to scroll01:01:16
through this a little bit01:01:22
so there's a bunch of others01:01:26
can you are you seeing that in the chat01:01:28
brian because yeah it's kind of01:01:29
difficult for me to01:01:31
determine what's the best approach to01:01:33
looking at this01:01:35
yeah so got that and the all active01:01:38
scenario legs had to be used01:01:43
yeah the customer equipment has to have01:01:46
a lag configured on it in an all active01:01:48
scenario to prevent loops from broadcast01:01:52
or01:01:53
multicast traffic01:01:56
so let me pull up slide01:01:59
here01:02:00
there it is01:02:03
and i don't want that01:02:08
darn message i should have stayed home01:02:09
my setup i'm more so used to it01:02:20
so in the event of a multicast traffic01:02:24
for example01:02:27
i can't use my pin here01:02:30
if i get a multicast frame for example01:02:32
coming into pe one or a broadcast frame01:02:35
coming into pe1 and i send it to ce101:02:39
ce1 could send that frame rate back01:02:43
into the pe and then we would be01:02:45
forwarding out to everybody and we'd01:02:47
have this loop set up01:02:49
so we01:02:50
need to have this lag created on here01:02:55
so that when it receives the multicast01:02:57
frame it's not going to send it out01:02:59
another interface associated to that leg01:03:10
get rid of that01:03:18
yeah it's the ce that needs the leg in01:03:21
there01:03:27
exactly yeah so it looks like01:03:31
jay answered that perfect01:03:36
what control plane types type 3 writes01:03:38
on that mpls it's an ingress replication01:03:41
point to multiple composite tunnels yeah01:03:44
so he answered that way to go jake01:03:47
uh when multiple communities added an01:03:50
export policy statement so we can use01:03:52
vrf import vrf export and vprn to column01:03:56
yes01:04:01
do we see01:04:03
so i'm assuming here i've got a question01:04:07
about01:04:08
the import and export policies for vrf01:04:11
targets and you are quite correct yes we01:04:14
have01:04:15
the ability to support multiple vrf01:04:18
targets in a vprn scenario01:04:20
i've not seen it configured in a vpls01:04:23
scenario but i have no reason to doubt01:04:26
that we could do so01:04:28
in our lab setups we're just using the01:04:32
standard01:04:33
vrf target target and then put in the01:04:36
community string target and the01:04:39
as number and unique identifier01:04:44
we are using01:04:46
uh01:04:48
vfrs01:04:50
so it's going to use targeted vfrs01:04:54
vrs i think is what we're saying here01:04:59
yes we're using route targets01:05:01
for all of our services whether it be a01:05:05
virtual private wire service or a01:05:07
virtual private land service or vprn01:05:10
service we are using route targets01:05:13
to identify which service01:05:15
will import which routes because01:05:17
remember this is uh01:05:20
bgp sessions are appearing to all the01:05:22
pes not all the pes are participating in01:05:25
the same services01:05:27
so we have to use these route targets to01:05:29
identify which services on which pes01:05:32
will be accepting the roads01:05:38
in the symmetric model they use01:05:43
nlri type 5 route01:05:46
pvpn01:05:48
yes01:05:50
i am no artist01:05:53
that is true01:05:55
yeah we are using the type 5 evpn routes01:05:58
in the symmetric model as well as you01:06:01
saw with the01:06:02
uh01:06:06
we also utilize in the symmetric models01:06:09
i go back01:06:10
let's go down here01:06:19
we have the options in the submetric01:06:21
model all three of the options we'll use01:06:24
the type five routes01:06:26
it's only in the interface full01:06:28
options whether they be numbered or not01:06:31
that we also use not just a type 5 but a01:06:35
type 2.01:06:37
so in this option01:06:39
only use the type 5 rep and this is what01:06:42
everybody's familiar with01:06:44
okay is the easiest one to put up it's01:06:47
interfaceless the vprns are populating01:06:50
the routes using the route targets we01:06:52
don't have to worry about another vpls01:06:56
service01:06:58
it's when we get into the interface full01:07:02
or we introduce this concept of the01:07:04
supplementary broadcast domain and now01:07:06
we're01:07:07
sending two routes01:07:09
to identify01:07:11
how to get to the gateway ip address01:07:14
as well as01:07:15
exchanging the01:07:17
ip prefix information01:07:25
now vprn is not a root rudder vpls01:07:28
routed vpls01:07:30
is a service that has a logical01:07:34
connection to a layer 3 service so we01:07:37
eroded vpls01:07:39
we don't have to01:07:41
if you had a hair color of mine you01:07:43
remember way back01:07:46
that in the01:07:47
service routers we used to either have01:07:49
to have an s hook that brought the01:07:52
bpls service to the vprn service01:07:55
or we had a versatile01:08:00
what was that versatile services module01:08:02
where we could do a logical connection01:08:05
but then we ident we created the routed01:08:07
vpls where we can just do a logical01:08:10
connection01:08:11
between the two services so a routed01:08:14
vpls is one with just a logical01:08:16
connection to a layer 3 service01:08:21
i used to have allow does must be that01:08:24
rvpls with allow ip internal binding01:08:28
and that's the way it used to be yes you01:08:31
used to have to put that allow internal01:08:33
ip binding01:08:35
but in the01:08:36
last couple releases they did away with01:08:39
that and we'll see that here when i01:08:41
create one01:08:45
and then we01:08:47
really don't need to identify that allow01:08:49
ip internal binding anymore01:08:55
i'll show you when i build one here01:09:00
okay01:09:02
that's we've got here01:09:06
and darren's gonna share the recording01:09:13
yeah i actually already shared it in the01:09:15
chat but i will send an email as well01:09:17
and hopefully get that at the next01:09:19
couple hours01:09:22
okay and we have gregory01:09:25
if you use the class a address space01:09:29
for example the ten network is a primary01:09:31
example of private ip address space01:09:35
uh01:09:36
then the saving of the ip address is01:09:38
unnecessary you are 100 correct yes01:09:42
okay you're not going to be advertising01:09:44
that01:09:45
ip address space like i said the 1001:09:47
network as you see here outside of your01:09:50
network there's all sorts of policies01:09:52
that will01:09:53
be deployed to make sure that any of the01:09:55
ip address spaces that are associated01:09:58
with the rfc 191801:10:00
will not leave your network so yes you01:10:03
are right01:10:10
okay i'm trying to contrast the layer 301:10:13
evpn with a01:10:14
traditional ip mpls vprn01:10:20
the additional interface seems more01:10:22
complicated than simply using the tunnel01:10:24
autobind yes it is01:10:27
is there another benefit for the01:10:28
removable ldp is a requirement the01:10:31
traditional model01:10:32
well that's that's really the only01:10:34
benefit is removing uh01:10:39
the requirement for another control01:10:42
plane protocol01:10:44
i don't need ldp i don't need rsvp in01:10:48
the control plane01:10:50
if i don't have a if i have a simple ip01:10:53
network then i can just simply do01:10:56
vxlan across the networks01:11:00
so that's the whole point about going to01:11:03
that01:11:05
but01:11:06
i don't see why we couldn't just do this01:11:08
as well inside that vxlan network i just01:11:11
need a vxlan tunnel and vni01:11:13
so this isn't a requirement01:11:17
to go across a vxlan network01:11:19
this is just an option01:11:23
okay the concept of a backhaul vpls was01:11:27
introduced years ago into the data01:11:29
centers01:11:31
and01:11:33
so this is uh01:11:34
just to carry on with those01:11:41
yes the auto bind in my standard vprn01:11:44
service01:11:53
and so yeah so it is a bit01:11:56
more complicated as for sure01:12:02
what else01:12:13
no you there was no requirement to01:12:15
attend an extra session gregory01:12:20
yesterday's session was strictly the sr01:12:23
linux session01:12:24
totally different box01:12:26
different cli01:12:30
and it was geared towards a data center01:12:37
that's what we got01:12:56
brian i wonder if we should uh go ahead01:12:58
with a demonstration i think there's01:13:00
still a lot01:13:01
i'm just worried that we we keep people01:13:03
on even though we still will share the01:13:05
recording we'll try to come back to the01:13:06
questions but i think uh01:13:08
i think it's best to me maybe just carry01:13:10
on right now all right we'll get going01:13:12
on this yes i don't want to do what i01:13:14
did yesterday01:13:17
so right now i'm in the customer01:13:18
equipment01:13:20
get into the session we're going to be01:13:21
doing a01:13:25
layer 2 service to interconnect ce5 to01:13:29
ce6 and right now as you can see i01:13:31
cannot ping ce5 to ce6 so01:13:36
it's going to fail and fail miserably01:13:40
so the first thing we have to do is take01:13:42
a look at our configurations01:13:51
as you see i already have bgp ebpn set01:13:55
up to pe 3 on this side of my data01:13:58
center it is established but we're not01:14:00
exchanging any routes because we don't01:14:02
have any services01:14:04
if i go to pe 301:14:13
i have my session back to pe one01:14:17
and i also have a session to pe for01:14:20
all both of these sessions are for01:14:23
address family evpn01:14:30
so the first thing i'm going to do is go01:14:32
on to pe1 and we're going to configure01:14:34
that vpls service01:14:44
because it is mdcli we have to do this01:14:49
get into the edit config mode so01:14:51
configure service we're going to do01:14:52
vpls101:14:55
i'm gonna turn it on01:14:58
we're gonna co01:15:00
if most of you know we have to associate01:15:02
a service with a customer now we're01:15:04
gonna start with the vxlan configuration01:15:09
we have got to give it an instance01:15:11
identifier and it's in here01:15:14
if i can spell we identify what we're01:15:16
going to use for a01:15:18
vxlan network identifier in this case01:15:21
we're doing01:15:23
bni101:15:30
then01:15:31
what we know we're doing bgp dash ebpn01:15:36
are01:15:37
evpn instance identifier01:15:40
is going to be one as well01:15:44
we're gonna for bgp vxlan one01:15:49
we're gonna enable that and we're going01:15:51
to associate it to that vxlan instance01:15:54
that i configured just a second ago01:15:56
which was vxlan instance one01:15:59
so now bgp evpn knows what service we01:16:02
have01:16:03
and because i associated it to vxlan01:16:06
instance one it knows what vni to use01:16:10
for this specific service01:16:17
i already did that01:16:20
okay01:16:30
now we have to connect01:16:33
ce1 or ce5 into this service and if we01:16:37
go back to ce501:16:39
we take a look at the interface it was01:16:41
using vlan tag one01:16:44
so i have to identify that vlan tag on01:16:47
my sap01:16:53
since it's model driven cli nothing's01:16:56
happened until i hit commit01:17:00
now with any luck01:17:03
oh yes i'm inside the configure mode01:17:13
so now you can see that we have sent01:17:16
one update01:17:18
that's because i turned on a vpls01:17:20
service01:17:22
it's using bgp evpn01:17:24
so that update i sent is my imet route01:17:28
the type three wrote to identify we have01:17:31
a service01:17:32
associated01:17:34
on pe one01:17:40
okay01:17:43
now we haven't received anything for pe301:17:45
and if i run this command again over on01:17:48
pe 3 we'll see that it doesn't have01:17:51
anything either01:17:52
because there is no service on pe 3 that01:17:56
matches the route target that we're01:17:57
utilizing from pe101:18:00
so now we gotta configure up pe301:18:04
so pe3's configuration01:18:11
we're going to go on to configure01:18:14
service again vpls1 in this instance01:18:21
still requires a customer01:18:24
so we'll use the default one01:18:35
so it's one we're to identify our v and01:18:38
i again01:18:40
but here here's where it gets a little01:18:42
different01:18:45
because we're going01:18:47
one side for vxlan the other side is01:18:49
going to be mpls we're going to need two01:18:52
instances of bgp so we have bgp instance01:18:55
one and they require two distinct route01:18:58
distinguishers01:19:07
so we'll use rote distinguisher here for01:19:09
bgp instance one01:19:11
and then we'll do another bgp instance01:19:14
only if i can spell01:19:28
let's just keep those routes different01:19:32
okay01:19:33
now we go back to bgp in this case evpn01:19:39
again we identify the evi service here01:19:43
that pgp will be advertising for our vpn01:19:46
again we're going to the vxlan now this01:19:49
one is identifying that the vxlan is01:19:52
going to be associated to that first bgp01:19:55
instance bgp instance one01:19:59
i'm going to turn this on01:20:02
and well we're going to relate this01:20:06
to that vxlan instance that i created01:20:09
earlier so it's going to be01:20:11
using the vni one01:20:18
then we01:20:20
get out of vx land01:20:23
stay in bgp01:20:25
we're going to do our mpls and associate01:20:27
it to that second01:20:30
instance of bgp01:20:35
uh we're gonna do an autobind01:20:37
because everybody loves autobind01:20:40
and we set up the resolution01:20:43
to use our filter01:20:45
and then we configure our filter01:20:48
for what kind of tunneling we're going01:20:50
to support in this case we're just going01:20:52
to support ldt01:20:56
and we exit out of this exit of this01:21:01
exit of this we do a quick info make01:21:03
sure i haven't missed anything01:21:07
so01:21:09
our service is enabled we got our vxlan01:21:12
instance one using vni one01:21:16
we created two bgp instances gave them a01:21:20
distinct route distinguisher for each01:21:22
one01:21:24
we associated our01:21:27
to bgp instance one01:21:31
and our mpls configuration we associated01:21:35
the bgp instance two01:21:39
under our vxlan for bgp vpn we referred01:21:43
it back to vxlan instance 1 which gives01:21:46
me my vni number01:21:49
and for mpls we did the standard01:21:52
autobind telling it to use an ldp tunnel01:21:58
so that looks like we got everything so01:22:00
if i do a commit01:22:04
and we do that01:22:08
quick up arrow we do a show bgp01:22:12
we now see that01:22:13
i not only have i sent one01:22:16
i've received one and i'm utilizing it01:22:18
so now we have know that pe 301:22:21
has sent out his imet wrote01:22:26
i keep forgetting01:22:27
this01:22:36
mdcli okay01:22:39
so now i've sent two01:22:42
here01:22:45
bgp so if i go over to pe401:22:50
one of the issues that we have here01:22:56
we go over to pe4 which is already01:22:59
pre-configured and we do a show service01:23:03
id1 which is what we're all configured01:23:05
with01:23:06
and i say what are my vxlan destinations01:23:09
for this service01:23:12
and you can see it identifies pe201:23:16
as a vx land destination but it's also01:23:18
identifying pe3 as a vx land destination01:23:23
and that's because we sent that update01:23:26
pe3 sent that update01:23:29
for vxlan and01:23:32
or mpls to both pe1 and pe401:23:36
because it's a bgp update and pe fork01:23:38
accepted it because they all have the01:23:40
same route target and it's all good01:23:44
so we know pe4 and pe3 are not connected01:23:47
via vx lan01:23:50
so we need to lay on a couple policies01:23:53
it's bgp bgp is controlled by policies01:23:58
okay so we've got a couple policies01:24:00
already configured01:24:03
uh exit all01:24:08
and if so01:24:11
and these things will get rid of them01:24:13
one is to reject it so if we go01:24:16
configure01:24:21
routine01:24:22
options policy01:24:24
now configure oh man01:24:28
ah there we go policy options01:24:34
so i've created a couple policies01:24:38
we've identified the community strings01:24:40
that are used for mpls and the community01:24:43
strings that are identifying the01:24:44
encapsulation for vxlan that pgp uses01:24:48
and then i created one called mpls01:24:51
underscore01:24:52
export01:24:54
where we're matching on evpn and the01:24:57
community string that identifies vx land01:25:00
communications and we will reject that01:25:04
and accept everything else01:25:06
and then for policy vx land export01:25:10
we're doing bgp evpn matching on the01:25:13
community string01:25:15
mpls01:25:19
and we're going to reject that and we01:25:22
send it off01:25:24
so01:25:38
so we have our two groups01:25:40
we have our group bvpn we have our two01:25:42
neighbors01:25:44
so a neighbor01:26:08
oh01:26:09
policy value what was that01:26:12
darn i forgot the policy names already01:26:20
mpls01:26:26
exlan export that's what i want01:26:34
there01:26:38
and then 10 to 10.10.401:26:52
there we go01:27:00
now01:27:01
that should01:27:06
taking a little while01:27:15
it should remove01:27:16
if i put the policies on correctly01:27:20
should remove01:27:24
and pls export so it should be doing01:27:26
that vxlan export chat that's right01:27:33
there just took bgp a little while01:27:36
so now pe4 no longer sees it01:27:41
so now01:27:45
you can send the ping01:27:51
we start getting the information01:27:56
and if we do a show01:27:58
router01:28:00
bgp01:28:02
summary again01:28:05
we're seeing that i've sent01:28:07
two ropes now01:28:10
one will be the imet road of course and01:28:12
the other one is the mac ip road01:28:14
identifying the mac information01:28:18
and if we do the same thing here01:28:27
again i'm sending out multiple routes01:28:32
with the mac ip information01:28:39
so if i do a on pe one01:28:52
we see the mac address01:28:55
that is associated to my mpls tunnel01:28:58
that goes over to01:29:00
router number four01:29:02
we learned it by evpn01:29:05
and the mac address that's using vxlan01:29:08
that goes to router number one01:29:12
so we see that we have the connectivity01:29:14
from vxlan01:29:16
through mpls01:29:18
back over to pe4 here01:29:34
he received he has an mpls01:29:36
infrastructure and then going over to01:29:38
pe2 he uses vxlan so went vxlan through01:29:42
mpls to vxlan01:29:45
to get our communications01:29:49
okay so we have layer two01:29:53
vpls service going across the network01:30:06
so that's it for the layer two stuff01:30:21
all right01:30:22
so01:30:24
we're going to take a look now01:30:26
because one of the common questions that01:30:28
we always get when we're teaching a01:30:30
class01:30:31
is01:30:32
most popular question that we get when01:30:34
we're teaching the class is what time01:30:35
does class end on friday01:30:37
but the second most popular question is01:30:40
what if01:30:41
so what if the core01:30:43
is not an evpn core01:30:47
so on pe 301:30:49
we do a show router01:30:52
bgp01:30:54
summary01:30:56
we see that we have an evpn session01:30:59
established to pe101:31:02
but to pe401:31:04
we're running vpn ipv4 so we have an01:31:07
mpls core01:31:09
that is not01:31:11
does not support evpn addressing01:31:14
so we're going to go straight across a01:31:16
standard ipv ip mpls core01:31:22
so on pe101:31:29
we have a number of services set up01:31:32
if we remember back to our diagram01:31:36
not this one01:31:41
uh oh actually was that one our diagram01:31:45
we have vpls11 that connects my ce01:31:48
device ce501:31:51
okay01:31:53
that vpls connect 11 is connected up01:31:56
into my vprn so let's take a look in01:31:59
there01:32:01
edit config private and we're going to01:32:04
go under configure service vpls 1101:32:08
because there was a question about this01:32:10
routed vpls01:32:13
as you can see we don't have to do that01:32:16
allow ip internal binding anymore with01:32:19
the newer releases all we i do is01:32:21
identify it to route a vpls in the01:32:24
service01:32:25
turn give it a customer and a sap and01:32:27
it's on01:32:32
if i go into vprn01:32:34
10001:32:38
within there i've got an interface to01:32:40
vpls1101:32:42
it identifies the service right here and01:32:44
that's the service name because that's01:32:46
what i configured up01:32:48
and then i have the gateway ip address01:32:51
for the ce devices right here01:32:54
i have the other interface that goes to01:32:56
my supplementary broadcast domain01:33:00
just have a mac address associated to it01:33:02
because i don't need an ip01:33:05
and we identify that we're using evpn01:33:08
tunnel for my s01:33:09
pd here connectivity01:33:14
so it's become quite a bit easier to01:33:16
make that routed vpls service so now on01:33:19
pe 301:33:21
if i do a show service01:33:24
service using01:33:26
say we don't have any so we have to01:33:28
configure these services01:33:30
on pe 3 to allow01:33:32
the ce5 to communicate to ce601:33:38
so the first thing we're going to01:33:39
configure up is the supplementary01:33:42
broadcast domain01:33:43
okay so we're going to go into edit01:33:46
config private01:33:48
and we're going to configure service01:33:50
vpls1001:33:54
customer01:33:55
one01:33:56
okay so we need since it's communicating01:33:59
to via vxlan01:34:01
similar to what we did before01:34:04
instance one01:34:06
we're going to give it a bni of 10 this01:34:08
time to match the other side01:34:16
we have to remember have to tell it that01:34:18
it is a road oh we got to back up though01:34:23
it is eroded bpls01:34:25
we still have to configure the bgpe vpn01:34:33
get out of the road of vpls bgp01:34:37
vpn01:34:39
evi in this case is 10 for my service01:34:43
identifier01:34:45
routes we're going to to import ip01:34:47
prefix and advertise those as true01:34:52
okay01:34:53
and then01:34:55
vxlan information01:34:59
in state enable01:35:01
or associate that back to vxlan instance01:35:04
one that i created01:35:09
and we01:35:10
exit out of that01:35:13
so that01:35:14
you do a quick info01:35:19
the only thing i haven't done is turned01:35:22
on the service yet01:35:24
but i said01:35:26
that looks like everything there so01:35:27
admin state enable01:35:32
and then commit01:35:35
all right so that looks good01:35:40
now if i do01:35:50
we now see that we're exchanging evpn01:35:53
information i sent one that'll be my01:35:55
imet wrote01:35:57
i've received01:35:59
a couple from the bgp from pe1 one of01:36:03
them be the imet ropes as well01:36:06
but as you see there's still nothing01:36:08
across my vpn01:36:12
so we'll exit out of this01:36:18
so now we got to configure up our vprn01:36:20
service on pe301:36:30
okay so now we're gonna enable it right01:36:33
now because sometimes i forget that give01:36:36
a customer one01:36:39
so we know01:36:41
that we're gonna do bgp ibpn01:36:44
because that's how we get to01:36:46
router number four we have mpls01:36:49
information here so we're gonna do do01:36:51
in-state enable on that01:36:55
we need a route distinguisher01:37:03
and we also need a vrf target01:37:06
and in this case we're going to just put01:37:08
in the community01:37:13
target01:37:22
okay01:37:23
so this command just identifies that i'm01:37:25
going to use that community string for01:37:28
both import and export rope targets01:37:33
we're going to do auto bind01:37:38
through our resolution fill by via01:37:40
filter01:37:41
and here we're going to just identify01:37:43
that the filter01:37:47
we'll be using ldp01:37:56
i do quick info make sure i got01:37:58
everything01:37:59
wrote distinguisher wrote target01:38:02
autobind01:38:04
that looks good01:38:12
just go back01:38:13
i need an interface01:38:16
to my spd01:38:19
only01:38:22
i'm going to give it a mac address01:38:37
free01:38:38
okay01:38:40
the vpls01:38:43
service that i'm going to bind it to is01:38:45
1001:38:46
evpn dash tunnel01:38:53
and that should be it01:38:56
so hopefully01:39:07
so i've got my service01:39:10
we know we're going across an iep01:39:14
core ipvpn cores standard mp uh ip mpls01:39:18
core01:39:19
i've identified the route distinguishers01:39:22
in the community01:39:24
we're going to use ldp to go between pe301:39:27
and pe01:39:28
401:39:29
and i've got the interface that connects01:39:31
me to my supplementary broadcast domain01:39:36
everything looks like i've got it01:39:37
enabled01:39:46
now01:39:48
if i go over to pe101:39:57
as long as i did everything correctly01:40:01
i see i've got the 11 network which is01:40:04
local to me01:40:05
and i've got the 12 network01:40:08
that is the one that is on hanging off01:40:10
of01:40:11
cr's pe201:40:15
okay so if we go over to pe201:40:18
which is on the other data center01:40:28
we'll see pe201:40:30
has the local network of pe1 in his01:40:33
forwarding database01:40:35
and as you can see the 12 network is01:40:38
local to pe 2.01:40:43
so now01:40:46
i go into my ce01:40:53
i see you have my of my local ip01:41:11
i have to source this because if i go01:41:13
outside of my01:41:15
local networks01:41:18
the router as most of you know the01:41:20
router will use the system ip address as01:41:22
its source01:41:24
and nobody over in pe2 knows how to get01:41:28
to that system ip address01:41:31
and as you see we are now pinging across01:41:34
the network01:41:36
i apologize folks for taking so long01:41:40
i know we've01:41:42
gone quite a bit over the time01:41:46
well maybe not no they're just coming up01:41:48
oh yeah we're just coming up yeah it's01:41:50
nine o'clock it was not the early start01:41:53
today was only a seven o'clock start01:41:56
there's potentially a lot more questions01:41:58
as well so we still could go over yeah01:42:01
let's01:42:02
let's hit the questions01:42:07
let's take over01:42:14
i got to remember to bring a mousepad01:42:16
when i come into these conference rooms01:42:21
yeah the trackers don't work so well on01:42:23
the uh unless you have a pad if it's a01:42:25
white table or something like that i01:42:27
don't think they work too well though01:42:29
i'm always moving this thing around01:42:31
trying to find it01:42:35
yes let's see we got a number of01:42:37
questions01:42:44
what is it here01:42:46
got me back in 10 minutes after the01:42:48
break so it came back01:42:52
still need stps though for mvpls and01:42:56
miri01:42:58
or the management vpls yet because the01:43:01
management vpls01:43:03
has to run the spanning tree through01:43:06
for01:43:07
one or more vpls services01:43:10
and so01:43:11
you would still need that sort of thing01:43:13
i've not actually tried to run a01:43:16
management vpls through vx well vxlan01:43:19
doesn't have loops because we're01:43:22
routing so it would only be if you're01:43:25
using mesh sdps and a combination of01:43:28
mesh and spoke sdps then i would need01:43:31
some sort of management vpls01:43:35
but yeah01:43:45
see here's the video from yesterday01:43:50
multiple communities yeah we talked01:43:53
about that01:43:57
in which sros is vpn's fully supported01:44:04
i don't know when the evpn support first01:44:07
came into the isros01:44:11
release01:44:13
i would imagine it was maybe a year or01:44:16
so ago since we just put the course up01:44:18
on it but01:44:21
i01:44:22
don't know the actual release number01:44:23
where first became fully integrated01:44:27
now all our courses have migrated to01:44:30
mdcli01:44:32
uh we still have one more to do i think01:44:36
jose is in the process of updating the01:44:39
multicast course to mdcli but i'm pretty01:44:42
sure everything else has been migrated01:44:45
to there i know i did the vpls course01:44:49
okay01:44:52
scaling differences between normal vpls01:44:55
and an evpn01:45:00
uh01:45:03
scaling is always an issue to discuss01:45:08
not so much from the service id numbers01:45:10
the id number values i don't think have01:45:13
changed01:45:15
scaling is always dependent upon01:45:19
more upon quality of service than it is01:45:21
the actual service itself how many cues01:45:24
how much memory01:45:26
is being consumed by each01:45:29
service and that's where a lot of the01:45:31
scaling issues come into is01:45:33
how much01:45:34
memory is each service consuming01:45:38
if you're not putting any queue01:45:40
structures in then you can probably01:45:42
support a lot more services01:45:45
so scaling numbers are always hard to01:45:48
quantify01:45:52
see01:45:54
in an all active scenario does the lag01:45:56
require the support of lacp protocol01:45:58
also01:45:59
no you really don't require it01:46:03
we can figure it up without lacp01:46:08
you can utilize that01:46:15
says take a screenshot with a snip tool01:46:17
to the new01:46:18
legs01:46:21
thanks right01:46:26
[Laughter]01:46:29
uh he's asking his leg01:46:32
two just simple legs01:46:35
yeah we as i said we we can just have a01:46:38
simple like we don't require the lacp01:46:40
packets01:46:42
uh the slide material01:46:44
will be offered up i believe darren's01:46:47
correct me if i'm wrong but it'll be01:46:49
offered up in a pdf format01:46:52
um01:46:53
it's a possibility i'll have to look um01:46:55
i i go on vacation next week so i can01:46:58
look at the pdfs i know there's some01:46:59
requests there there'll definitely be a01:47:01
recording01:47:02
um i just need to get it uploaded but um01:47:05
i'll see if we can also do the slides as01:47:07
well01:47:14
yeah the lacp i see one here from jay01:47:18
for the legs lacp for the eggs you would01:47:20
require01:47:22
and he's correct in identifying that if01:47:24
there was some spoofing01:47:27
okay to prevent spoofing from somebody01:47:29
trying to say that they are a system01:47:32
id and they really aren't01:47:37
reporting session when i miss parts01:47:39
thanks01:47:41
cross connect aggregate group oh jason01:47:44
[Laughter]01:47:46
yes01:47:47
the versatile services module with the01:47:50
crossconnect aggregate group she had the01:47:52
a-pass and the b-pass01:48:05
oh we asked the01:48:07
understand the bgp peerings i actually01:48:10
had the bgp was actually01:48:13
uh configured earlier so let me get into01:48:17
here pe101:48:22
get out of there01:48:32
so when we configure up bgp i put in a01:48:35
group01:48:36
i identified the as number and that01:48:40
we're using family evpn so we did have01:48:43
to do it in the base routing bgp context01:48:48
okay01:48:49
so we're the we already pre-configured01:48:52
these routers01:48:53
with the appropriate contacts here if we01:48:59
out uh here we go on to config01:49:02
router01:49:03
bgp01:49:06
you'll see that i have two families01:49:08
one that has vpn ipv4 and one for evpn01:49:13
and we associated pe1 with the evpn01:49:16
family01:49:17
and pe 401:49:19
with the vpn ipv4 family01:49:23
sorry i prob i should have gone through01:49:25
that01:49:35
well in our case what is the underlay in01:49:37
terms of igp in our case yes we are01:49:40
using an igp as the underlay um01:49:45
it's not a requirement anymore it used01:49:47
to be that you could not use bgp to01:49:50
resolve a bgp next top01:49:53
so i initial in the older versions of01:49:56
bgp i couldn't use ebgp to resolve an01:50:00
ibgp next top but now you can there is a01:50:02
switch that you can turn on01:50:05
that will allow bgp to use bgp to01:50:08
resolve the next hump but in this case01:50:11
we are doing uh01:50:14
if i do a show route again we do have an01:50:17
igp running right now01:50:22
and we're using ospf okay but uh01:50:30
does the vni need to be the same as the01:50:33
evi no it does not it just needs to be01:50:35
unique okay ever so that's all we need01:50:39
from it it needs to be unique01:50:46
what if we have evpn vxlan towards pe101:50:51
tells you what if was the second best01:50:53
question01:50:54
and evpn vxlan between pe401:50:58
do we need to create two bgp instances01:51:01
well that's very similar01:51:07
to this but no in this case no you01:51:10
wouldn't need to create two bgp01:51:12
instances01:51:14
because you would then be able to send01:51:16
that vxlan flooding information across01:51:19
so we'd be able to if all of them are01:51:22
part of the same services so we could01:51:24
just go ahead with a single vb bgp01:51:27
instances01:51:33
the route target is auto-generated yes01:51:36
it utilizes the uh if we take a look01:51:48
here we do uh01:52:03
let's do an01:52:05
imat01:52:08
inclusive there we go01:52:15
and01:52:21
there we go so01:52:22
as we see here01:52:25
we created target and it was01:52:27
auto-generated using the autonomous01:52:30
system number01:52:32
and in this case since we have vpls with01:52:34
an evi number one it used the evi number01:52:38
to auto create the community so the a s01:52:41
and evi number were used to auto-create01:52:44
the target01:52:59
now01:53:00
the two bgp what's the purpose of the01:53:02
dual bgp instances one and two and that01:53:05
was just to01:53:07
have a separate instance that would01:53:09
handle vxlan and the hand another second01:53:12
instance that would handle the mpls01:53:15
side as far as the advertisement goes01:53:19
so they both serve the same purpose01:53:22
but across different transport protocols01:53:41
yes mdcli has been quite a learning01:53:45
experiences01:53:48
is there a free 7750 image for learning01:53:51
purposes that i do not know tom01:53:55
i don't know01:53:57
that would be something01:54:00
i'd have to look and see if there is01:54:02
such a thing01:54:14
okay virtual service runner okay good01:54:17
thank you kevin01:54:26
yeah we uh01:54:28
as far as training labs yes darren01:54:30
talked about it earlier today there is a01:54:33
training knives my sr labs where you can01:54:36
book time on it and you can identify01:54:39
various01:54:40
src scenarios that you want to populate01:54:43
i believe it's01:54:44
three hours at a crack01:54:48
is there a vsr image that will give you01:54:51
that you can just load up on your01:54:53
unit or linux box01:54:55
i don't know01:54:57
i really don't know about that one01:55:06
yeah we do support shout out to01:55:08
container lab yes01:55:14
what is the uses of bgp ebpn and bgp01:55:18
vpls where was that01:55:29
pgp evpn01:55:32
is to01:55:34
basically put all the parameters that we01:55:36
want bgp01:55:37
the bgp based context to advertise on01:55:40
our behalf01:55:43
pgp vpls01:55:50
[Music]01:55:52
and that would be this one01:56:05
oh wrong wrong router i'm on the wrong01:56:07
way01:56:18
so this wouldn't be a good one to go01:56:20
let's uh log out all these01:56:24
let's get this guy this is one's just01:56:26
doing bpls01:56:44
i'm not seeing bgp vpls anywhere01:56:52
here let's take a look01:56:56
well it is there01:56:59
let's see what we got01:57:02
in groups01:57:03
mid-state applied groups01:57:07
maximum v i oh yes01:57:12
yes yes yes this bgp vpls scenario let's01:57:15
see if i can remember01:57:17
this is for01:57:19
auto configurations you would01:57:22
identify01:57:24
a label block01:57:27
for this vpls service and then01:57:31
you would automatically01:57:33
create the service labels it's been a01:57:35
while since i did this but you would use01:57:38
this block and there's a calculation it01:57:40
would go through01:57:42
to automatically generate or identify01:57:45
what service label you're going to use01:57:47
to get to this vpls service01:57:50
so instead of advertising out01:57:53
like using uh tldp to advertise a01:57:56
service label you would just strictly01:57:58
use bgp vpls to advertise a block of01:58:01
labels01:58:02
and then the remote pe can then go01:58:05
through this calculation to identify01:58:08
which01:58:10
label you're using for this specific01:58:12
service01:58:14
yeah it's covered in our vpls course01:58:17
i have not taught it in a couple years01:58:20
so i'm a little fuzzy on it but as01:58:23
soon as i identified the max v e i d and01:58:26
v i v e numbers01:58:29
then that's where it comes from i'd i'll01:58:31
have to get a01:58:33
go through the course or for our vpls01:58:35
course to01:58:36
get a better understanding of it again01:58:39
but that's what it's used for so we01:58:41
don't need to have01:58:43
tldp to identify what your service01:58:46
labels we can just advertise out a block01:58:48
of labels01:58:49
and then there's a calculation that the01:58:51
remote pe does to identify what is the01:58:54
service label for this specific service01:59:02
uh01:59:04
does this example work just as well with01:59:06
ipv601:59:10
oh01:59:11
ian dickinson what are we talking about01:59:14
here my friend01:59:15
example are we just talking about01:59:18
the services themselves then yes if it's01:59:21
a vprn ipv4 vpr and ipv6 it would work01:59:30
yeah01:59:31
oh you're talking about the proxy01:59:33
then yes for neighbor discovery cache01:59:36
rather than arp yes that would work01:59:51
share the sample configuration01:59:55
this configuration here i could i could01:59:57
share that with you i guess02:00:00
it's the same thing you would get out of02:00:02
our courseware a little different i02:00:04
modified it a bit same sort of thing02:00:17
is our image02:00:21
and now it's02:00:23
looks like we're coming to the end of02:00:25
the questions02:00:30
because everything i'm seeing is people02:00:32
saying they're leaving02:00:35
yeah i think we'll probably wrap it up02:00:36
now um thanks a lot everyone for02:00:38
attending02:00:41
there is a link i put in here for02:00:42
evaluation so if you get a chance please02:00:44
take the opportunity to provide an02:00:46
evaluation02:00:47
and i also put a link for our linkedin02:00:49
group as well02:00:51
maybe02:00:52
i'll put both of these links in again02:00:54
since it may have got lost in the02:00:56
chat so there's the linkedin group so02:01:02
and then uh02:01:04
here's the link for the evaluation as02:01:06
well02:01:09
so again thanks a lot everyone for02:01:10
attending uh we hope you uh got some02:01:12
value of this presentation02:01:14
and uh02:01:16
thanks a lot brian for taking the time02:01:17
and uh everyone have yourself a great02:01:19
rest your day02:01:20
and i'll get it may not it may be a02:01:22
while before i get the recording for02:01:24
this just only because i'm on vacation02:01:27
starting tomorrow yeah i think your02:01:29
vacation is more important there oh my02:01:32
goodness02:01:34
hey we don't get a lot of warm weather02:01:36
in canada so you got to take advantage02:01:38
while you can02:01:40
i i too would like to extend my02:01:42
appreciation to you all for the02:01:44
questions and the attention that you02:01:46
spentDescription:
In this webinar, we'll discuss the concepts and benefits of L2 and L3 EVPN services. We'll also discuss at a high level the various BGP EVPN route types used to support the EVPN services, the basics of a L2 EVPN service, and its support for multi-homing. Finally, we’ll discuss the different interface-ful numbered forwarding model for L3 EVPN services. The discussion will be followed by a demonstration of the configuration of an L2 EVPN service crossing multiple data centers using VXLAN / MPLS interconnect. Learn more about the Nokia Service Routing Certification (SRC) Program - https://www.nokia.com/networks/training/src/
Preparing download options
* — If the video is playing in a new tab, go to it, then right-click on the video and select "Save video as..."
** — Link intended for online playback in specialized players
Questions about downloading video
How can I download "Introduction to L2/L3 EVPN Services Supported on the Nokia 7750 Service Router" video?
http://unidownloader.com/ website is the best way to download a video or a separate audio track if you want to do without installing programs and extensions.
The UDL Helper extension is a convenient button that is seamlessly integrated into YouTube, Instagram and OK.ru sites for fast content download.
UDL Client program (for Windows) is the most powerful solution that supports more than 900 websites, social networks and video hosting sites, as well as any video quality that is available in the source.
UDL Lite is a really convenient way to access a website from your mobile device. With its help, you can easily download videos directly to your smartphone.
Which format of "Introduction to L2/L3 EVPN Services Supported on the Nokia 7750 Service Router" video should I choose?
The best quality formats are FullHD (1080p), 2K (1440p), 4K (2160p) and 8K (4320p). The higher the resolution of your screen, the higher the video quality should be. However, there are other factors to consider: download speed, amount of free space, and device performance during playback.
Why does my computer freeze when loading a "Introduction to L2/L3 EVPN Services Supported on the Nokia 7750 Service Router" video?
The browser/computer should not freeze completely! If this happens, please report it with a link to the video. Sometimes videos cannot be downloaded directly in a suitable format, so we have added the ability to convert the file to the desired format. In some cases, this process may actively use computer resources.
How can I download "Introduction to L2/L3 EVPN Services Supported on the Nokia 7750 Service Router" video to my phone?
You can download a video to your smartphone using the website or the PWA application UDL Lite. It is also possible to send a download link via QR code using the UDL Helper extension.
How can I download an audio track (music) to MP3 "Introduction to L2/L3 EVPN Services Supported on the Nokia 7750 Service Router"?
The most convenient way is to use the UDL Client program, which supports converting video to MP3 format. In some cases, MP3 can also be downloaded through the UDL Helper extension.
How can I save a frame from a video "Introduction to L2/L3 EVPN Services Supported on the Nokia 7750 Service Router"?
This feature is available in the UDL Helper extension. Make sure that "Show the video snapshot button" is checked in the settings. A camera icon should appear in the lower right corner of the player to the left of the "Settings" icon. When you click on it, the current frame from the video will be saved to your computer in JPEG format.
What's the price of all this stuff?
It costs nothing. Our services are absolutely free for all users. There are no PRO subscriptions, no restrictions on the number or maximum length of downloaded videos.