Simple explanation
Networks fail — the question is whether that failure is invisible (high availability working as intended) or catastrophic (no plan in place). This lesson covers the concepts and metrics behind designing for resilience.
Technical explanation
- Redundancy — eliminating single points of failure through duplicate hardware, links, or paths.
- Failover — automatically switching to a backup component/path when the primary fails, ideally with minimal or no service interruption.
- Active-active vs. active-passive — active-active runs multiple redundant components simultaneously handling load together; active-passive keeps a backup idle until needed.
- Site types: hot site (fully operational, near-instant failover, expensive), warm site (partially configured, faster than cold but not instant), cold site (basic infrastructure only, requires significant setup time before use).
- Metrics: RTO (Recovery Time Objective — how long you can tolerate being down), RPO (Recovery Point Objective — how much data loss, measured in time, is acceptable), MTTR (Mean Time To Repair), MTBF (Mean Time Between Failures — a reliability metric).
- NIC teaming — combining multiple network interface cards for redundancy and/or increased throughput at the server level.
- First Hop Redundancy Protocols (FHRP): HSRP and VRRP — allow multiple routers to share a virtual IP address, so if the primary router fails, a backup transparently takes over as the default gateway without hosts needing reconfiguration.
Synonyms / related terms
| Term | Means | |---|---| | FHRP | First Hop Redundancy Protocol | | HSRP | Hot Standby Router Protocol (Cisco) | | VRRP | Virtual Router Redundancy Protocol (open standard) |
Concept Check
"A company can tolerate up to 4 hours of downtime after a disaster, but can only afford to lose 15 minutes of data." These are two different metrics being described: the 4-hour tolerance is RTO, and the 15-minute data loss tolerance is RPO — a common exam trap is conflating the two, but they answer fundamentally different questions (how long can we be down vs. how much data can we lose).
Interview-style Q&A
Q: Why would an organization choose a warm site over a hot site, given the faster recovery a hot site offers? A: "Cost. A hot site is essentially a fully duplicated, constantly-running environment — expensive to maintain indefinitely for a disaster that may never happen. A warm site is a middle ground: enough pre-positioned infrastructure to recover faster than a cold site, at a fraction of a hot site's ongoing cost. The choice comes down to how tight the actual RTO requirement is versus what the organization is willing to spend to meet it."
Memory trick
"RTO is Time, RPO is Data" — a two-word anchor for the single most commonly confused pair of terms in this whole domain.