Three-tier topology · Series/parallel reliability · Fail links & devices, watch availability
This simulator models the availability and connectivity of a three-tier campus network (core, distribution, access) as links and devices are failed interactively. Network engineers use it to visualize how redundant paths affect overall uptime and to understand which single points of failure will cause complete connectivity loss.
Each network tier is a series element: the end-to-end availability is the product of all individual tier availabilities (A_total = A1 × A2 × A3 × ...). When a redundant path (Path B) is added at a tier, that tier becomes a parallel block whose combined availability is 1 − (1 − A_A)(1 − A_B), which is dramatically higher than either path alone. For example, two 99.9% paths in parallel yield 99.9999% availability.
Connectivity is lost only when every path through any single tier is simultaneously down — this is the definition of a complete network outage. The simulator computes per-tier availability and multiplies them for the system total, then expresses the result as downtime minutes per year and a "number of nines" metric.
IEEE 802.1D (Spanning Tree Protocol) and 802.1w (RSTP) govern loop prevention in redundant switched networks, with RSTP providing sub-second convergence versus STP's 30–50 second convergence. IEEE 802.3ad (LACP) provides link aggregation for active-active uplink bonding. First-hop redundancy protocols HSRP (Cisco) and VRRP (RFC 5798) provide gateway redundancy. TIA-942 recommends redundant network infrastructure (dual power feeds, dual uplinks) for Tier 2 and above data center designs. Uptime Institute Tier classifications map to availability requirements: Tier III requires 99.982% (1.6 hr/yr downtime).
The access layer is typically a single point of failure unless dual-homed using LACP or stacking. Adding redundant uplinks at the distribution and core tiers provides the greatest improvement in system availability per dollar spent. Spanning tree root bridge placement should be the core layer to prevent suboptimal traffic paths when redundant paths are active. For voice and video, failover time is critical: RSTP converges in under 1 second, while LACP failover is sub-100 ms, making link aggregation preferable where supported. Always account for PSU redundancy within each device, as internal power failure is a leading cause of unplanned outages.
Start with the default three-tier topology and observe the system availability with all devices up. Enable redundant paths (Path B) at the core and distribution tiers to see how availability improves. Then click "Primary (A) failed" on individual tiers to simulate a device or link failure and watch whether connectivity is maintained via the redundant path. Remove redundant paths to reveal which tiers create single points of failure. The availability results panel shows the system figure, downtime, and number of nines in real time.
Number of nines is -log10(1 - A) where A is fractional availability. 99.9% = 3 nines (8.76 hr/yr downtime), 99.99% = 4 nines (52.6 min/yr), 99.999% = 5 nines (5.26 min/yr). Enterprise networks typically target 4–5 nines for critical infrastructure.
Legacy 802.1D STP has a 30–50 second convergence time due to listening and learning timer delays. RSTP (802.1w) uses proposal/agreement handshakes to achieve sub-second convergence on point-to-point links, making it suitable for voice and video networks where any outage exceeding 150 ms causes perceptible quality degradation.
LACP (IEEE 802.3ad) provides active-active link aggregation with sub-100 ms failover and doubled bandwidth between two devices. Use LACP for server uplinks and core-to-distribution uplinks where both bandwidth and resilience are needed. STP handles loop prevention across the broader switched domain where LACP cannot be used.
Enterprise-grade core switches (Cisco Catalyst 9500, Juniper EX9200) typically achieve 99.999% hardware availability with dual PSUs and hot-swap supervisors. ISP circuits are typically rated at 99.9–99.95% by SLA, making the WAN circuit the most common weak link in end-to-end availability.
Yes in theory, but management complexity introduces operational risk. Misconfigured spanning tree or incorrect LACP settings can cause outages worse than a non-redundant design. Redundancy must be paired with proper configuration management, change control, and testing.
Try our Enterprise IT Networks Studio
More calculators, simulators, and guides for this discipline.