Skip to content
Log in Sign up My account

1.1 Enterprise Design and High Availability

Blueprint 1.1: Explain different design principles used in an enterprise network

  • 1.1.a High-level enterprise network design such as 2-tier, 3-tier, fabric, and cloud
  • 1.1.b High availability techniques such as redundancy, FHRP, and SSO

Hierarchical campus design

The hierarchical model splits the campus into layers, each with one job. That makes the network predictable to scale, troubleshoot and summarise.

Layer Job Typical features
Access Connect endpoints VLANs, PoE, port security, 802.1X, QoS trust boundary, STP edge (PortFast, BPDU Guard)
Distribution Aggregate access, policy boundary Default gateway with FHRP, routing to core, summarisation, ACLs, L2/L3 boundary
Core Fast transport between distribution blocks High-speed L3 links, no policy, minimal features, fast convergence

Three-tier

Access, distribution and a separate core. Use it when there are several distribution blocks (multiple buildings, data centre, WAN edge). Without a core, connecting n distribution blocks in a full mesh needs n(n-1)/2 inter-block link pairs, which stops scaling after three or four blocks.

Two-tier (collapsed core)

Distribution and core are combined in one pair of switches. Right for a single building or small campus where a separate core adds cost and no benefit.

graph TB
  subgraph "Three-tier"
    C1[Core 1] --- C2[Core 2]
    D1[Dist A1] --- C1 & C2
    D2[Dist A2] --- C1 & C2
    D3[Dist B1] --- C1 & C2
    D4[Dist B2] --- C1 & C2
    AA[Access A] --- D1 & D2
    AB[Access B] --- D3 & D4
  end
  subgraph "Two-tier collapsed core"
    CD1[Core/Dist 1] --- CD2[Core/Dist 2]
    AC[Access] --- CD1 & CD2
  end

Layer 2 vs Layer 3 access

L2 access (traditional) Routed access
Access-to-distribution links 802.1Q trunks Routed point-to-point links
Default gateway Distribution, needs FHRP Access switch SVI, no FHRP needed
STP Active, blocks redundant uplinks Confined to the access switch
Convergence STP and FHRP timers Routing protocol (sub-second with tuning)
VLAN spanning access switches Possible Not possible without an overlay

Fabric (spine-leaf and SD-Access)

A fabric separates a simple routed underlay (IP reachability between loopbacks) from a virtual overlay (VXLAN tunnels carrying user traffic). Spine-leaf is the data-centre form: every leaf connects to every spine, so any leaf to any leaf is always two hops with ECMP across all spines. SD-Access is the campus form, covered in 1.3.

Cloud

Model You manage Provider manages Example
IaaS OS, apps, data Hardware, virtualisation Catalyst 8000v in AWS or Azure
PaaS Apps, data Runtime, OS, hardware Managed app platforms
SaaS Data and access Everything else Microsoft 365, Webex

Deployment models: public, private (on-premises), hybrid (private plus public with connectivity between them) and multicloud. Design concerns for the exam: latency to the cloud, where internet breakout happens (centralised vs direct internet access at the branch), security inspection in the path, and bandwidth cost.

High availability

Availability comes from removing single points of failure at each level, then making failover fast.

graph LR
  A[Redundant links<br/>EtherChannel, ECMP] --> B[Redundant gateways<br/>HSRP, VRRP, GLBP]
  B --> C[Redundant supervisors<br/>SSO + NSF]
  C --> D[Redundant chassis<br/>StackWise Virtual, VSS]

Redundancy

  • Link: EtherChannel bundles links so one failure does not trigger STP. Equal-cost routed links let the routing protocol load-share and fail over.
  • Device: pairs of distribution and core switches. StackWise Virtual (and the older VSS) joins two chassis into one logical switch so access switches can use multichassis EtherChannel with no blocked ports.
  • Path: dual WAN or ISP links with a routing protocol or IP SLA tracking to choose and fail over.

First Hop Redundancy Protocols

Hosts have one default gateway. An FHRP lets two or more routers share a virtual IP and MAC so the gateway survives a router failure.

HSRP VRRP GLBP
Standard Cisco IETF (RFC 5798 for v3) Cisco
Roles Active / Standby Master / Backup AVG / AVF
Load balancing Per group (multiple groups) Per group Native, per host, across up to 4 AVFs
Virtual MAC v1 0000.0c07.acXX, v2 0000.0c9f.fXXX 0000.5e00.01XX 0007.b400.XXYY
Preempt default Off On Off (AVG); AVF preempt on
Multicast address v1 224.0.0.2, v2 224.0.0.102 224.0.0.18 224.0.0.102
Real IP as virtual IP No Yes (owner gets priority 255) No

Configuration lives in ENCOR 3.3.c; for 1.1 you need to explain why FHRP exists and how it fits the design (distribution layer, aligned with the STP root).

Exam trap

Align the HSRP active router with the STP root bridge for that VLAN. If they differ, traffic crosses the inter-distribution link on every frame. The exam likes this as a "suboptimal path" scenario.

SSO and NSF

A chassis with two route processors or supervisors can fail over between them.

  • SSO (Stateful Switchover): the standby supervisor is kept synchronised with the active one (configuration, interface state, L2 tables). On failover, the standby takes over without resetting line cards, so forwarding continues. SSO alone does not preserve the routing protocol sessions' state.
  • NSF (Non-Stop Forwarding): keeps forwarding using the last known FIB while the new active supervisor rebuilds routing tables. It uses graceful restart so neighbours keep routes instead of tearing down the adjacency.
  • An NSF-capable router is the one restarting. An NSF-aware neighbour helps by holding routes during the restart. Most current IOS XE routers are NSF-aware by default.
  • NSR (Nonstop Routing) is the alternative: the standby keeps full protocol state so neighbours don't even need to be NSF-aware.
redundancy
 mode sso
!
show redundancy
show redundancy states

router ospf 1
 nsf ietf          ! RFC 3623 graceful restart
! or: nsf cisco    ! Cisco proprietary
!
show ip ospf | include Non-Stop|restart

router bgp 65001
 bgp graceful-restart
!
show ip bgp neighbors 192.0.2.1 | include Graceful

Exam trap

SSO plus NSF is the pair to remember. SSO keeps the box forwarding through a supervisor failover; NSF stops the neighbours from dropping routes during it. FHRP is the answer when the whole router fails, not just a supervisor.

Lab: build and fail a two-tier block

Topology: base topology, ACC1, DIST1, DIST2. Goal: see each HA layer at work.

  1. Trunk ACC1 to both distribution switches. Create VLAN 10.
  2. Make DIST1 the STP root for VLAN 10 and HSRP active (priority 110, preempt). DIST2 is secondary root and standby.
  3. Bundle DIST1 to DIST2 with a two-link LACP EtherChannel.
  4. From Host1, run a continuous ping to an address beyond the distribution pair.
  5. Shut one member of the EtherChannel. Expected: no lost pings.
  6. Shut DIST1 entirely. Expected: a few lost pings while HSRP fails over (default hello 3 s, hold 10 s). Repeat with standby 10 timers msec 200 msec 750 and compare.
! DIST1
spanning-tree vlan 10 root primary
interface Vlan10
 ip address 10.10.10.2 255.255.255.0
 standby version 2
 standby 10 ip 10.10.10.1
 standby 10 priority 110
 standby 10 preempt
!
show standby brief
show spanning-tree vlan 10 | include root|Root
show etherchannel summary

Quick check

Why does a three-tier design scale better than a fully meshed two-tier design across many buildings?

The core gives each distribution block one pair of uplinks to the core. Without it, every distribution pair must connect to every other, which grows with the square of the number of blocks.

A supervisor fails on a modular switch running SSO but not NSF. What happens to OSPF?

The switch keeps forwarding in hardware, but the new active supervisor must re-form OSPF adjacencies. Neighbours detect the adjacency reset and withdraw routes through the switch until it reconverges. NSF (graceful restart) prevents that.

References

  • Cisco Campus LAN design guides: search "Campus LAN and Wireless LAN Solution Design Guide" on cisco.com (CVD)
  • Cisco Live On-Demand Library: search "campus design" and "high availability"
  • RFC 5798 (VRRPv3), RFC 3623 (OSPF graceful restart), RFC 4724 (BGP graceful restart)