burningnode.com
Random networking related stuff.

Redundancy vs load-balancing


This post stands as a simple introduction to redundancy and load balancing. These terms are often associated and mixed up, but are two different things.

What is redundancy?
  • Several devices working “together“
  • Failover: in case of incident the backup device take the load

Why redundancy? 

  • Need to improve availability
  • Ensure availability of critical applications and systems (C.I.A Triad)
  • Ensure customer satisfaction
  • Avoid trust and financial loss

How to achieve redundancy? 

  • Use of redundancy and load balancing protocols (HSRP, CARP, VRRP)
What is load balancing?
  • Several devices working “together”
  • Load sharing: every device processes packets

Why load balancing?

  • Need to improve availability and traffic processing capacity
  • Ensure high availability of critical systems
  • Ensure customer satisfaction
  • Avoid trust and financial loss

How to achieve load balancing?

  • Use of load balancing protocols (GLBP, CARP)
  • Use advanced configurations of redundancy protocol (MHSRP)

How it is working?

Redundancy:
  • One device considered “active” and the other(s) “passive”
  • Only the “active” device processes the packets
  • The passive device is waiting for the failure.
  • Simulate a virtual interface with a virtual IP / MAC address.
Load balancing:
  • All devices are active.
  • All devices process the packets.
  • Load is divided among the devices.
  • Simulate a virtual interface with an IP address (MAC address are used for load balancing purpose).

 Redundancy and load balancing principles

Redundancy LB
Redundancy scenario YES NO
Load balancing scenario YES YES

There is also a slight difference between load balancing and load sharing: load-balancing implies an equal load distribution between the nodes whereas load sharing is more rough and implies that every gateway in the group processes packets.