Architecture · Topic 2 of 5
High Availability, FHRP and SSO
Learn high availability for ENCOR: redundant devices and links, first hop redundancy protocols, and SSO stateful switchover on dual supervisor switches.
- 5 min
- High availability removes single points of failure by adding redundancy at several levels.
- An FHRP gives hosts a virtual gateway address that survives the loss of one router.
- HSRP and GLBP are Cisco protocols. VRRP is an open standard.
What it is
High availability means the network keeps working when a device, a link or a part of a device fails.
No single technique gives it. You combine redundancy at each level: links, devices, gateways and the control plane inside one chassis. This page covers three of the main techniques, which are plain redundancy, first hop redundancy protocols and stateful switchover.
How it works
Each technique protects a different point of failure.
Redundancy of links and devices
The basic idea is to avoid a single point of failure. You install two of everything that matters and connect them so either can carry the traffic.
- Link redundancy: two uplinks from each access switch, often combined in an EtherChannel or kept loop free by Spanning Tree.
- Device redundancy: two distribution switches, two core switches, two routers at the WAN edge.
- Path redundancy: a routing protocol finds another route when the first one fails.
- Power and supervisor redundancy: two power supplies, and in some chassis switches two supervisor modules.
Redundancy must be tested. A backup path that has never carried traffic may hide a configuration mistake.
First hop redundancy protocols
A host has one default gateway address. If that router fails, the host cannot leave its subnet, even if a second router is available.
A first hop redundancy protocol (FHRP) solves this. Two or more routers share a virtual IP address and a virtual MAC address, and hosts use the virtual address as the gateway.
One router answers for the virtual address. If it fails, another takes over, and the hosts change nothing.
| Protocol | Standard | Notes |
|---|---|---|
| HSRP | Cisco | One active router, one standby, the others listen |
| VRRP | Open standard (IETF) | One master router, the others are backups |
| GLBP | Cisco | Several routers can forward at once, so it can share load |
HSRP and VRRP have their own configuration page in the Infrastructure chapter. For this topic you need to know the purpose and the roles.
Stateful switchover (SSO)
Some chassis have two supervisor modules, which are the control plane cards. One is active and one is standby.
In SSO mode the standby supervisor keeps a copy of the active supervisor's state. If the active one fails, the standby takes over quickly. Forwarding in hardware can continue during the change, so users see little or no loss.
The related feature Nonstop Forwarding (NSF) lets the device keep forwarding while its routing protocols recover. Neighbours that support graceful restart keep their routes during that time. SSO and NSF are used together.
You can check the redundancy state on a platform that has redundant supervisors:
Switch(config)# redundancy
Switch(config-red)# mode sso
Switch(config-red)# end
Switch# show redundancyIn the output, look for the operating redundancy mode and the state of the two supervisors. A healthy pair shows the mode as sso, one supervisor active and the other in a standby hot state.
Redundancy modes compared
Older redundancy modes exist on some platforms, and they recover more slowly.
| Mode | Idea | Recovery |
|---|---|---|
| SSO | Standby holds synchronised state | Fast, forwarding continues |
| RPR | Standby is loaded but not fully synchronised | Slower |
Support for these modes depends on the platform and software release, so check the hardware guide for the device you use.
Stacking and virtual chassis
Some designs join physical switches into one logical switch. A stack or a virtual switching system presents two or more boxes as a single device with one control plane. Neighbours see one switch, so the design may need less FHRP configuration, depending on the platform and layout.
What to remember for the exam
Seven facts cover most questions on this topic.
- High availability removes single points of failure by adding redundancy at several levels.
- An FHRP gives hosts a virtual gateway address that survives the loss of one router.
- HSRP and GLBP are Cisco protocols. VRRP is an open standard.
- GLBP can load balance across several gateways. HSRP and VRRP use one forwarder at a time.
- SSO keeps the standby supervisor synchronised with the active one.
- NSF lets forwarding continue while routing protocols reconverge.
show redundancyshows the redundancy mode and supervisor states.
If a question says the hosts need no change when the gateway fails, the answer is an FHRP, because the virtual address does not move.
Common mistakes
These four mistakes appear often.
Calling FHRP a routing protocol. It does not exchange routes. It only decides which router answers for the virtual gateway address.
Assuming SSO is the same as having two switches. SSO is about two supervisors in one chassis, not two separate devices.
Believing redundancy works without testing. A standby that was never exercised may fail when needed.
Forgetting NSF needs neighbour support. Graceful restart works only when the neighbours understand it.
Do not assume that every Catalyst switch has dual supervisors. SSO applies to platforms that have redundant control plane hardware.
Check yourself
Work through these three questions.
What problem does an FHRP solve for a host?
A host has only one default gateway address, and an FHRP gives it a virtual address that stays reachable if one router fails.
Which FHRP is an open standard?
VRRP is the open standard, while HSRP and GLBP are Cisco protocols.
What does SSO do when the active supervisor fails?
The standby supervisor already holds the synchronised state, so it takes over quickly and forwarding can continue.
Topic 2 of 5 in Architecture
Test yourself on Architecture
The full Cisco CCNP 350-401 ENCOR question bank has 854 questions, with timed practice exams and an explanation for every answer.
This study guide is independent and is not affiliated with, endorsed by, or sponsored by Cisco. Vendor names, certification names and exam codes are trademarks of their respective owners and are used for identification only.