Posts

VXLAN Journey - Part 13 - Multi-Site VXLAN Anycast Border Gateway - fabric-advertise-pip

Image
In part - 12 , we have changed the next-hop announced by anycast border-gateways over the dci-links (inter-site). In this blog we will change the next-hop announced by anycast bgws over fabric-links (intra-site) from vip (loopback 100) to pip (loopback 1). This change will be valid for type-5 routes only ; type-2 routes towards the site-internal vteps will still be announced with vip as next-hop. I am neutral in my recommendation to enable this feature. Enablement of this feature requires change in all the switches in in the fabric - Bgw, Spine and Leaf. Disclaimer: The feature increases the memory consumption in Leaf and Spine switches because BGP's rib and routing tables need to install more routes. So, enable this feature cautiously following the Nexus switch's hardware scalability guide . This feature is only valid for type-5 routes only . Our topology looks below -  01 - Network Topology We have Vrf_Blue (l3vni - 100000) which is extended over multisite. Then vlan 102 (l2...
Post a Comment
Read more

VXLAN Journey - Part 12 - Multi-Site VXLAN Anycast Border Gateway - dci-advertise-pip

Image
In previous two blogs; part - 10 and part - 11 , we have extended our data-centers Dc-01 and Dc-02 using anycast bgws. The bgws were announcing the type-2 (mac and mac+ip) and type-5 (prefix) routes across dci-links using their shared vip (loopback 100) as next-hop. In this blog we will change that behavior. The bgws will announce type-2 and type-5 routes using their PIP (loopback 1) as next-hop across dci-links. This feature is called - " dci-advertise-pip ". With this we can achieve more utilization of our dci-links using ecmp - because now traffic is load-shared using individual pip address; it is not load-shared using the vip address. My recommendation is to enable this feature in anycast border-gateways. Disclaimer: The feature increases the memory consumption in the border-gateways because BGP's rib and routing tables need to install more routes. So, enable this feature cautiously following the Nexus switch's hardware scalability guide .  Let's start con...
Post a Comment
Read more

VXLAN Journey - Part 11 - Multi-Site VXLAN Anycast Border Gateway Mac-Vrf (L2VNI) extension

Image
In part - 10 we have configured Ip-Vrf (l3vni) extension between our data-centers using anycast border-gateways. Now we will configure Mac-Vrf (l2vni) extesion between our data-centers. What is multi-site ? When we connect two different dc-fabrics (different BGP ASN) together with eBGP l2vpn-evpn - we call it vxlan multi-site extension. What is l2vni extension ? If vlan - A, under vrf - X, from Site - 1 can communicate over the border-gateways with vlan - A, under vrf - X  from Site - 2;  we call it l2vni/mac-vrf extension over multisite. The common thing in the extension is that communicating vlans are the same . Our topology looks like below -  01 - Network Topology We are going to extend l2vni - 10010 (vlan 101) so that Vlan-101-Pc-01 and  Vlan-101-Pc-02 can communicate with each other using the border-gateways. For Spine, Leaf and Border-Gateway configurations have a look in the previous blogs. We will only show new configurations required of l2vni extensio...
Post a Comment
Read more