Cisco - Understanding High Availability¶

Leveraging Logical Switches¶

Switches as each network layer should be deployed in pairs for redundancy
Traditional devices provided redudancy but don’t allow all links to be used
Redudancy only provided at core and distribution layers, not access
Logical switches permits grouping the redudant links into an etherchannel, no blocked links
Two approaches to logical switches
- StackWise
- Virtual Switching System (VSS)

StackWise and StackWise Plus options available
Supported on Catalyst 3750-E, 3750-X and 3850
Uses special “stacking” cables deployed in a daisy-chain to form a ring
Traffic is carried bi-directional across the stacking switch fabric
Switches can be added/removed as long the ring is kept intact in one or both directions
One switch is appointed as the master to perform management functions
Single management IP is assigned to the entire switch stack

FHRP protocols (VRRP, HSRP, GLBP) provide HA only for the default gateway IP, no assistane for directly connected hosts
Chassis switch can support multiple supervisors, offering redundancy within the switch itself
Some switches also have multiple power supplies offering the ability to support dual power feeds in the event of power failure

First of the two supervisors to boot becomes active, others take on a standby role
Active supervisor boots to a fully initialised and operational state
Standby only initialised to a certain level depending on configured/supported modes
- Route Processor Redudancy (RPR)
- Route Processor Reudancy Plus (RPR+)
- Stateful Switch Over (SSO) - Best
Single Router Mode (SRM) indicates two route processors but one one is active
Dual Router Mode (DRM) indicates both route processors are active, commonly alongside HSRP
SRM is not comatile with RPR/RPR+
SRM is inherent with SSO, can be seen as “SRM with SSO”
Route Processor Redundancy (RPR) Mode
- Supervisor only partially booted
- Every module in switch is reloaded when active fails
- Layer 2/Layer 3 not initialised
Route Processor Redundancy Plus (RPR) Mode
- Supervisor and Route Engine Initialised
- No Layer 2/Layer 3 functions started
- No reload of modules needed when active fails
- Switch ports retain existing state
Stateful Switch Over (SSO) Mode
- Supervisor Fully Booted And Initialised
- Startup/Running config synchronised
- Layer 2 information maintained
- No change to interface states
- Layer 3 protocols and routing protocol convergence happens upon active failure

Set the redundancy mode for the chassis

NOTES:

Needs to be configured on both switches the first time it is setup
To support RPR+ both chassis must have identical IOS otherwise fails back to RPR

redundancy
  mode { rpr | rpr-plus | sso }

Verify Redundancy Status

show redundancy status

Configure Suprvisor Synchronisation

NOTE: By default startup and running configurations are sync’d

redudancy
  main-cpu
    auto-sync standard # default setting
    auto-sync { startup-config | config-register | bootvar }

Configure NSF For BGP

router bgp <as>
  bgp graceful-restart

Configure NSF For EIGRP

router eigrp <as>
  nsf

Configure NSF For OSPF

router ospf <process-id>
  nsf

Configure NSF For IS-IS

router isis [<tag>]
  nsf [cisco | ietf ]
  nsf interval [<mins>]
  nsf t3 { manual | [seconds] | <adjacency> }
  nsf interface wait <seconds>