Cisco Cisco Prime Network Services Controller 3.0 Livre blanc
© 2014 Cisco and/or its affiliates. All rights reserved. This document is Cisco Public Information.
Page 17 of 24
Logical Rules for Service Appliances Placement
As best practice for a predictable and stable VDC, Cisco recommends a smart placement for the network service
appliances in the certain PoD. It makes more sense to run them on dedicated hardware (a services cluster). Just
as the physical appliances run virtual instances and are placed in a common predefined and known place, this is
the only way to support QoS and guarantee SLAs for those services. Several vendors and customers now follow
that guidance.
On top of the smart placement, anti-affinity rules make sure High Availability (HA) is actually implemented and
virtual appliance HA pairs don’t end up running on the same host. This capability, which has long been built-in with
virtual appliance HA pairs don’t end up running on the same host. This capability, which has long been built-in with
physical appliances, now needs to be verified in the virtual data center. Making sure that a pair of network service
appliances doesn’t end up running on the same host guarantees device failover but more granular placement rules
appliances doesn’t end up running on the same host guarantees device failover but more granular placement rules
need to be implemented to make sure certain network service appliances (the non-hypervisor-based types) are
placed in a more robust cluster that can utilize traffic and device resources to guarantee QoS. The implementation
details for virtual services cluster and QoS is out of scope for this paper.
A stable cloud infrastructure design should include network service appliances smart placement (Figure 10).
Figure 10. Smart Placement of Virtual Appliances