22.05.2017 Views

nx.os.and.cisco.nexus.switching.2nd.edition.1587143046

Nexus Switching 2nd Edition

Nexus Switching 2nd Edition

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

(GLBP). In a spanned Layer 2 environment, it is important to retain the same defaultgateway<br />

IP <strong>and</strong> MAC address at each site to allow for seamless live migrations of virtual<br />

resources between sites. With HSRP deployed as the FHRP, the vMAC is derived from the<br />

HSRP group number. So, if data center 1 has HSRP group number x between the two routers<br />

<strong>and</strong> data center 2 has the same HSRP group number of X, both data center, will have the<br />

same vMAC programmed locally. With a server cluster that fails over or when a virtual<br />

machine moves from one data center to another (assuming all of the back-end storage is in<br />

place for this to happen), the mac-address of its default gateway is cached. There will not<br />

be any traffic disruption as the end h<strong>os</strong>t will have the same gateway vMAC address.<br />

Note<br />

HSRP <strong>and</strong> VRRP assign gateway virtual MAC addresses in a deterministic fashion.<br />

GLBP, by default, assigns virtual MAC addresses to the various gateways via the<br />

active virtual gateway (AVG). Because of this difference in behavior, FHRP<br />

localization currently requires that either HSRP or VRRP be deployed.<br />

This functionality causes the OTV AEDs to filter FHRP messages acr<strong>os</strong>s the overlay<br />

network to enable the co-existence of the default gateway VIP at each OTV site. Without<br />

filtering enabled, FHRP messages would be sent acr<strong>os</strong>s the overlay leading to the election<br />

of a single active device to serve as the default gateway. This behavior would then force<br />

any traffic destined for the default gateway to the site where the Active node was elected—<br />

leading to suboptimal traffic flow <strong>and</strong> also a high amount of data traversing the inter-DC<br />

links creating unnecessary use of th<strong>os</strong>e links. However, with filtering enabled, FHRP<br />

messages are dropped at the AED, <strong>and</strong> the gateway VIP can be considered active-active at<br />

each OTV site—thus eliminating hair-pinning of traffic destined for the default gateway<br />

between sites. In addition, with the HSRP filtering only the local routers in each data center<br />

will know about each other; the routers in the other data centers do not know anything about<br />

the others. This can result in active or active HSRP (with vPC in this sample topology)<br />

configuration <strong>and</strong> cannot have HSRP “listeners” as you filter all the HSRP Multicast hell<strong>os</strong>.<br />

FHRP filtering functionality should be deployed with an equivalent method to optimize<br />

inbound traffic flows to avoid asymmetric traffic behavior. (This would be unwanted<br />

especially in deployments leveraging stateful services such as firewalls or load balancers<br />

acr<strong>os</strong>s data centers.) The next section covers several methods for ingress routing<br />

optimization.<br />

Example 11-21 shows how to prevent HSRP messages to perform HSRP localization.<br />

Example 11-21. Define HSRPv1 <strong>and</strong> HSRPv2 to Block HSRP Hello Messages Between<br />

the OTV Edge Devices<br />

Click here to view code image<br />

ip access-list ALL_IPs<br />

10 permit ip any any

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!