Hi Gyan,

In-line, please.
Thanks,
Jorge

From: Gyan Mishra <[email protected]>
Date: Thursday, March 26, 2020 at 8:50 AM
To: "Rabadan, Jorge (Nokia - US/Mountain View)" <[email protected]>
Cc: BESS <[email protected]>, "UTTARO, JAMES" <[email protected]>
Subject: Re: [bess] vxlan evpn and L3 vpn interworking standard


Hi Jorge

Thank you for the correction I did mean type 2 so that’s good the host routes 
can be blocked.

What are your thoughts on the placement of the gateway or composite PE 
inter-working function on the core or DC side?
[Jorge] as you can see we wrote procedures for the gateway and composite cases, 
and even for the gateway+composite case. With those three cases I think we can 
address most use-cases. My thoughts? Today I mostly see the gateway use-case, 
where the gateway function is placed into the DCGW. Jim was also showing his 
use-case, that it uses gateways connecting the EVPN domains into the ipvpn one. 
But as more features are added to EVPN, I can see evpn and ipvpn PEs in the 
same domain and the composite function being used more and more.

Understood on eve BGP LU tunnel to stitch the domain together so it’s 
contiguous.  Could you also just extend and overlay the BGP EVPN AFI/SAFI over 
the MPLS core domains as an option for DCI.
[Jorge] yup

Thanks

Gyan

On Thu, Mar 26, 2020 at 3:22 AM Rabadan, Jorge (Nokia - US/Mountain View) 
<[email protected]<mailto:[email protected]>> wrote:
Hi Gyan,

“Is it possible in either gateway or composite PE interworking function that we 
can only export out the type 5 routes and block the type 4 host routes from 
being exported into IPVPN.  In most all cases there would be a L3 VNI tenant 
VRF prefix which would be represented by the Type 5 which is sufficient of 
reachability. “

I think you mean EVPN type 2 routes. Yes, you should be able to do that.

“I did notice at the bottom of page 6 & top of page 7 single domain scenario 
depicted.  I was curious to understand in that scenario how does BGP-LU come 
into play??”

Note that a “domain” is defined as follows: “Two PEs are in the same domain if 
they are attached to the same tenant and the packets between them do not 
require a data path IP lookup (in the tenant space) in any intermediate 
router.” In figure 3 we wanted to illustrate an example where PE1 and PE2 
belong to the same domain, no ip-lookups in between. For that, you need to 
connect PE1 and PE2 by a tunnel that can cross different ASNs, and BGP-LU can 
do that.

Hope it helps.
Thanks.
Jorge


From: BESS <[email protected]<mailto:[email protected]>> on behalf of 
Gyan Mishra <[email protected]<mailto:[email protected]>>
Date: Wednesday, March 25, 2020 at 10:44 PM
To: "UTTARO, JAMES" <[email protected]<mailto:[email protected]>>
Cc: BESS <[email protected]<mailto:[email protected]>>
Subject: Re: [bess] vxlan evpn and L3 vpn interworking standard

Hi Jim

I read through the draft.  This is exactly what I was looking for.  So in our 
scenario we would use the gateway PE option where we are exporting the EVPN 
routes & importing  VPNv4 & VPNv6 and vice versa and would use the "D Path" 
attribute for loop avoidance between domains.  Our DCI scenario would be like  
your drawing depicted where our Data Center core "router server" box that 
inter-connects all the PODs vxlan fabrics via BGP EVPN in the Data Center - 
This box would act as the demark between the "DC" and "Core" and would be the 
one doing the import/export "gateway PE" inter-working function to the IPVPN PE 
for core to data center reachability..

I noticed in section 4.3 the aggregation of routes option as we would not want 
the host routes to be propagated into IPVPN.

In the ineterworking option if we wanted to extend each tenant VRF natively 1-1 
mapping into unique L3 VPN VRF or take all the EVPN tenant VRFs ISF SAFI and 
import all into a single IPVPN VRF do we have that flexibility to map as 
desired.  Also if the Gateway PE was our DC route server box and we wanted to 
extend to the core natively  each tenant VRF 1-1 mapping into different unique 
IPVPN VRFs - since this is now an inter-AS LSP -exiting the DC to the PE the DC 
PE & Core PE would be acting as "inter-as" ASBRs in that case and would have to 
do an inter-as opt a,b,c,ab style.

I think the tricky part is where the Demark "Gateway PE" resides -- on the Core 
side or DC side.  If it ends up being the DC box the DC to Core is eBGP peering 
and then you would need to do inter-as L3 vpn option to extend the LSP. 
However, if the "Gateway PE" function resides on the Core PE then you would 
extend EVPN AFI to the core PE and then perform the gateway function where now 
its iBGP in the core and use can use the core RR for your LSP and now the edge 
inter-as LSP complexity does not occur.

Is it possible in either gateway or composite PE interworking function that we 
can only export out the type 5 routes and block the type 4 host routes from 
being exported into IPVPN.  In most all cases there would be a L3 VNI tenant 
VRF prefix which would be represented by the Type 5 which is sufficient of 
reachability.

Our other scenario for DCI interconnect would be extending EVPN AFI/SAFI over 
the mpls core.  This scenario keeps IPVPN & EVPN separate.

I did notice at the bottom of page 6 & top of page 7 single domain scenario 
depicted.  I was curious to understand in that scenario how does BGP-LU come 
into play??


<------------------------------------------------->

                                BGP-LU

           <------------------------------------------------->



                          ASBR------------ASBR

                         +------+       +------+

           +-------------|      |       |      |-------------+

          PE1            +------+       +--+---+            PE2

        +------+   DC1      |     WAN      |     DC2     +------+

    TS1-|IP-VRF|   EVPN     |              |     EVPN    |IP-VRF|-TS2

        +------+          ASBR            ASBR           +---+--+

           |             +------+       +------+             |

           +-------------|      |       |      |-------------+

                         +------+       +------+

                            +--------------+



           <--------------------DOMAIN-1--------------------->

Kind regards

Gyan
Verizon
Cell 301 502-1347

On Wed, Mar 25, 2020 at 8:16 AM UTTARO, JAMES 
<[email protected]<mailto:[email protected]>> wrote:
Gyan,

              Does the attached slide correlate with your requirement? My 
thoughts are that the interworking needs to support both intra WAN and WAN 
connectivity to DCs.. The introduction of D-PATH is required to ensure that a 
path p can traverse multiple signaling domains i.e EVPN<->L3VPN<->EVPN.. I 
would be interested in your use cases..

Thanks,
              Jim Uttaro

From: BESS <[email protected]<mailto:[email protected]>> On Behalf Of 
Gyan Mishra
Sent: Wednesday, March 25, 2020 1:37 AM
To: BESS <[email protected]<mailto:[email protected]>>
Subject: [bess] vxlan evpn and L3 vpn interworking standard


Dear BESS,

Does anyone know if this draft has been picked up in another draft or is on its 
way to become standardized.  Please provide data tracker if so.

This is a much needed feature for operators for DC for mpls core connectivity 
DCI extension.

https://datatracker.ietf.org/doc/draft-rabadan-sajassi-bess-evpn-ipvpn-interworking/<https://urldefense.proofpoint.com/v2/url?u=https-3A__datatracker.ietf.org_doc_draft-2Drabadan-2Dsajassi-2Dbess-2Devpn-2Dipvpn-2Dinterworking_&d=DwMFaQ&c=LFYZ-o9_HUMeMTSQicvjIg&r=s7ZzB4JbPv3nYuoSx5Gy8Q&m=Q5Qb9X0Q_fH2abLb-9fPBNqCki_KBGPuOaZeJsigPGA&s=g_SDL98eh20_aja820tdFnjy-_G79Zekfvd3t5YTEoQ&e=>

I see this draft is close to being published but is intra DC integration with 
IP mpls vpn.  I am not completely clear if the inter working function by the 
draft above is completely picked up by this draft.

https://tools.ietf..org/html/draft-ietf-bess-evpn-inter-subnet-forwarding-08<https://urldefense.proofpoint.com/v2/url?u=https-3A__tools.ietf.org_html_draft-2Dietf-2Dbess-2Devpn-2Dinter-2Dsubnet-2Dforwarding-2D08&d=DwMFaQ&c=LFYZ-o9_HUMeMTSQicvjIg&r=s7ZzB4JbPv3nYuoSx5Gy8Q&m=Q5Qb9X0Q_fH2abLb-9fPBNqCki_KBGPuOaZeJsigPGA&s=Bv-Oy0p0_u2EDBbejfuQYIoR8hOUinTQS0CF-Y_DpfU&e=>


Kind regards

Gyan
Verizon
cell 301 502-1347

Sent from my iPhone


--
Gyan  Mishra
Network Engineering & Technology
Verizon
Silver Spring, MD 20904
Phone: 301 502-1347
Email: [email protected]<mailto:[email protected]>


--
Gyan  Mishra
Network Engineering & Technology
Verizon
Silver Spring, MD 20904
Phone: 301 502-1347
Email: [email protected]<mailto:[email protected]>


_______________________________________________
BESS mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/bess

Reply via email to