Every formula, worked example, and free calculator for IP subnetting, CIDR, VLSM, bandwidth planning, latency, TCP/IP, VoIP, WiFi coverage, and 60+ more network engineering calculations — all in one place.
The most-used networking calculations — divide IP address space into logical subnets for routing, security, and efficiency.
Subnetting divides a large IP network into smaller, more manageable segments. Every subnet calculation comes down to two fundamental formulas:
Number of Subnets = 2^n (n = bits borrowed from host portion)
Hosts per Subnet = 2^h - 2 (h = remaining host bits; -2 for network + broadcast)
Subnet Mask = 32 - prefix length (e.g., /24 = 255.255.255.0)
Given: 192.168.1.0/24. You need 6 subnets.
Borrow 3 bits (2^3 = 8 subnets ≥ 6 needed). New prefix = /27. Each subnet has 2^5 - 2 = 30 usable hosts. Subnet increment = 256 - 224 = 32. Subnets: .0, .32, .64, .96, .128, .160, .192, .224.
| Class | Range | Default Mask | Networks | Hosts/Net |
|---|---|---|---|---|
| A | 1.0.0.0 – 126.255.255.255 | /8 | 126 | 16,777,214 |
| B | 128.0.0.0 – 191.255.255.255 | /16 | 16,384 | 65,534 |
| C | 192.0.0.0 – 223.255.255.255 | /24 | 2,097,152 | 254 |
| Private A | 10.0.0.0/8 | /8 | 1 | 16,777,214 |
| Private B | 172.16.0.0/12 | /12 | 16 | 65,534 |
| Private C | 192.168.0.0/16 | /16 | 256 | 254 |
| Prefix | Subnet Mask | Hosts | Use Case |
|---|---|---|---|
| /30 | 255.255.255.252 | 2 | Point-to-point WAN links |
| /29 | 255.255.255.248 | 6 | Small office segments |
| /28 | 255.255.255.240 | 14 | VLAN segments |
| /27 | 255.255.255.224 | 30 | Small floors/departments |
| /26 | 255.255.255.192 | 62 | Medium departments |
| /25 | 255.255.255.128 | 126 | Large departments |
| /24 | 255.255.255.0 | 254 | Standard LAN segment |
| /23 | 255.255.254.0 | 510 | Large LAN segments |
| /22 | 255.255.252.0 | 1,022 | Campus networks |
| /16 | 255.255.0.0 | 65,534 | Large enterprise |
Efficient address allocation using Classless Inter-Domain Routing and Variable Length Subnet Masking.
CIDR replaces classful addressing by expressing the network prefix directly in the address: 192.168.0.0/22 means the first 22 bits identify the network. This enables route aggregation (supernetting) and efficient address allocation that was impossible under Class A/B/C rules.
Total addresses = 2^(32 - prefix)
Usable hosts = 2^(32 - prefix) - 2
Supernet: find common bits between two networks, reduce prefix by 1 per aggregation
VLSM allows you to allocate different-sized subnets from the same address block. The golden rule: always allocate the largest subnet first, then work downward. This prevents address space fragmentation.
Example: You have 10.0.0.0/24 and need subnets for 60 hosts, 30 hosts, 12 hosts, and 2 WAN links. Allocate in order: /26 (62 hosts), /27 (30 hosts), /28 (14 hosts), /30 + /30 (2 hosts each). Total used: 64 + 32 + 16 + 4 + 4 = 120 addresses of 254 available.
Wildcard masks are the inverse of subnet masks, used in Cisco ACLs and OSPF. A 0-bit means "must match," a 1-bit means "any value is acceptable."
Wildcard Mask = 255.255.255.255 - Subnet Mask
Example: 255.255.255.0 -> Wildcard: 0.0.0.255 (matches entire /24)
Example: 255.255.255.252 -> Wildcard: 0.0.0.3 (matches /30 pair)
IPv6 uses 128-bit addresses with hexadecimal notation. A typical allocation: ISP assigns a /48, you create /64 subnets for LAN segments. Each /64 contains 18.4 quintillion addresses — enough that every device gets a unique global address.
| IPv6 Prefix | Addresses | Standard Use |
|---|---|---|
| /128 | 1 | Single host (loopback) |
| /64 | 18.4 quintillion | Standard LAN subnet |
| /48 | 65,536 x /64s | Typical enterprise allocation |
| /32 | 65,536 x /48s | Large ISP block |
Determine how much bandwidth your network actually needs — from individual links to full enterprise WAN planning.
These three terms are often confused but measure different things. Bandwidth is the theoretical maximum (e.g., 1 Gbps port speed). Throughput is actual data transferred per second — always less than bandwidth due to protocol overhead and errors. Goodput is useful application data only, excluding retransmissions and headers. Goodput is typically 85-95% of throughput.
Required BW = (Users x Avg. usage per user x Peak factor) / Link efficiency
Peak factor typically 1.5x – 3x average usage
Link efficiency = 0.7 to 0.8 (never plan for 100% utilization)
To calculate how long a file transfer takes across a given link:
Transfer Time (seconds) = File size (bits) / Bandwidth (bps)
Example: 1 GB file over 100 Mbps = (1 x 8 x 10^9) / (100 x 10^6) = 80 seconds
Add latency for WAN: Total = Transfer time + (RTT x number of round trips)
BDP tells you how much data can be "in flight" at any moment. It's critical for tuning TCP window sizes on high-bandwidth, high-latency paths (satellite, intercontinental).
BDP (bits) = Bandwidth (bps) x RTT (seconds)
Example: 1 Gbps link, 100ms RTT: BDP = 1,000,000,000 x 0.1 = 100,000,000 bits = 12.5 MB
TCP window must be >= BDP to fully utilize the link
| Application | Bandwidth per User | Notes |
|---|---|---|
| Web browsing | 1-2 Mbps | Average; bursts to 5+ Mbps |
| Video streaming HD | 5-8 Mbps | Netflix HD recommendation |
| Video streaming 4K | 15-25 Mbps | Netflix 4K recommendation |
| Video conferencing | 1.5-4 Mbps | Zoom HD video call |
| VoIP G.711 | 87 Kbps | Including headers |
| VoIP G.729 | 31 Kbps | Compressed, lower quality |
| Cloud backup | 10-100 Mbps | Burst during backup window |
Measure, diagnose, and optimize network delay — from RTT and jitter to packet loss impact on application performance.
Total network latency is the sum of four delay types. Understanding each one tells you where to optimize:
Total Latency = Propagation + Transmission + Processing + Queuing
Propagation delay = Distance / Signal speed (~200,000 km/s in fiber)
Transmission delay = Packet size (bits) / Link rate (bps)
RTT (round-trip time) = 2 x one-way latency
Jitter is the variation in packet arrival times. It severely impacts real-time applications like VoIP and video. Acceptable jitter is below 30ms for VoIP. Jitter = |Delay of packet N - Delay of packet N-1|. Jitter buffers compensate by 20-150ms at the cost of added latency.
Packet loss triggers TCP's congestion control, exponentially reducing throughput. The Mathis formula quantifies this:
Throughput = (MSS / RTT) x C / sqrt(loss_rate)
C = constant approximately 1.22
At 1% loss: throughput drops to ~12% of no-loss rate on a 100ms RTT link
| Metric | Excellent | Acceptable | Poor |
|---|---|---|---|
| Latency (LAN) | <1ms | <5ms | >20ms |
| Latency (WAN) | <20ms | <100ms | >200ms |
| Jitter | <5ms | <30ms | >50ms |
| Packet loss | <0.1% | <1% | >2.5% |
| RTT (VoIP) | <150ms | <300ms | >400ms |
Deep-dive calculations for TCP window sizing, MTU/MSS optimization, Ethernet frames, and protocol overhead.
TCP's receive window limits how much unacknowledged data can be in transit. On high-latency links, a small window throttles throughput far below the link's capacity.
Max TCP Throughput = Window Size (bytes) x 8 / RTT (seconds)
Example: 64KB window, 200ms RTT = (65,536 x 8) / 0.2 = 2.62 Mbps max
For 10 Gbps over 100ms RTT: Window needed = 10Gbps x 0.1s / 8 = 125 MB
MTU (Maximum Transmission Unit) is the largest frame a link can carry. On Ethernet, MTU = 1500 bytes. MSS (Maximum Segment Size) = MTU - IP header - TCP header = 1500 - 20 - 20 = 1460 bytes. Mismatched MTU causes black-hole routing and silent performance issues.
MSS = MTU - IP header (20B) - TCP header (20B) = MTU - 40 bytes
With IP options: MSS = MTU - 60 (max IP header) - 60 (max TCP header)
GRE tunnel: effective MTU = physical MTU - 24 bytes GRE overhead
IPSec tunnel (ESP): effective MTU = physical MTU - ~73 bytes
An Ethernet II frame contains: Preamble (8B) + Destination MAC (6B) + Source MAC (6B) + EtherType (2B) + Payload (46-1500B) + FCS (4B). Minimum frame = 64 bytes, maximum = 1518 bytes (without VLAN). Jumbo frames extend to 9000 bytes for data center use.
| Protocol | Header Size | Overhead % | Notes |
|---|---|---|---|
| IPv4 | 20-60 bytes | 1.3-4% | Options rarely used |
| IPv6 | 40 bytes fixed | 2.7% | No fragmentation at routers |
| TCP | 20-60 bytes | 1.3-4% | Options add overhead |
| UDP | 8 bytes | 0.5% | Minimal overhead |
| GRE | 24 bytes | 1.6% | Tunneling protocol |
| IPSec ESP | 50-73 bytes | 3.3-4.9% | Encryption overhead |
| MPLS | 4 bytes/label | 0.3%/label | Label stacking adds overhead |
Coverage planning, signal strength analysis, channel selection, and throughput estimation for 802.11 networks.
FSPL is the signal attenuation in open space — it increases with both distance and frequency. This is the baseline calculation before adding indoor attenuation factors.
FSPL (dB) = 20*log10(d) + 20*log10(f) + 92.4
where d = distance in km, f = frequency in GHz
5 GHz at 50m: FSPL = 20*log10(0.05) + 20*log10(5) + 92.4 = 66 dB
2.4 GHz at 50m: FSPL = 20*log10(0.05) + 20*log10(2.4) + 92.4 = 60 dB
| RSSI (dBm) | Signal Quality | Expected Performance |
|---|---|---|
| -30 to -50 | Excellent | Full speed, ideal for 4K streaming |
| -50 to -60 | Good | Reliable for video calls and streaming |
| -60 to -70 | Fair | Web browsing, basic tasks |
| -70 to -80 | Poor | Marginal connectivity, frequent drops |
| Below -80 | Unusable | Connection likely fails |
WiFi 6 (802.11ax) introduces OFDMA, MU-MIMO (8x8), and 1024-QAM modulation. Theoretical max throughput per band: WiFi 5 = 3.5 Gbps (5 GHz, 4x4 MIMO, 160 MHz). WiFi 6 = 9.6 Gbps (combined dual band). Real-world throughput is typically 40-60% of theoretical maximum.
Throughput (Mbps) = Modulation bits x Code rate x Subcarriers x Spatial streams x Guard interval efficiency
WiFi 6 max: 1024-QAM (10 bits) x 5/6 x 980 x 8 x 0.914 = 9607.8 Mbps theoretical
Accurately size bandwidth for VoIP deployments — from a single call to thousands of concurrent sessions.
VoIP bandwidth is often underestimated because people forget protocol headers. Each RTP packet carries voice payload plus UDP/IP/Ethernet headers. The total per-call bandwidth depends on codec choice and packetization interval.
Payload size = Codec bitrate x Packetization interval
Packet size = Payload + RTP(12B) + UDP(8B) + IP(20B) + Ethernet(18B)
Packets/sec = 1000ms / Packetization interval (ms)
Total BW = Packet size (bits) x Packets/sec
| Codec | Bitrate | Packet Int. | BW/Call (incl. headers) | Quality (MOS) |
|---|---|---|---|---|
| G.711 | 64 Kbps | 20ms | 87.2 Kbps | 4.1 (toll quality) |
| G.729 | 8 Kbps | 20ms | 31.2 Kbps | 3.92 |
| G.722 | 64 Kbps | 20ms | 87.2 Kbps | 4.5 (HD voice) |
| G.723.1 | 5.3/6.3 Kbps | 30ms | 21.9/24.0 Kbps | 3.65/3.9 |
| Opus | 6-510 Kbps | 20ms | Variable | Up to 4.5 |
| iLBC | 13.3/15.2 Kbps | 30/20ms | 27.0/29.0 Kbps | 4.14 |
OSPF cost, STP port cost, VLAN planning, DHCP scope design, BGP path analysis, and MAC address tools.
OSPF uses interface cost to determine the shortest path. Cisco's default reference bandwidth is 100 Mbps — this means FastEthernet and Gigabit Ethernet have the same cost (1) unless you change the reference bandwidth.
OSPF Cost = Reference Bandwidth / Interface Bandwidth
Default: Reference BW = 100 Mbps
GigabitEthernet: 100,000,000 / 1,000,000,000 = 1 (rounds down to 1)
Best practice: Set reference BW = 100 Gbps (100,000 Mbps) for modern networks
| Link Speed | 802.1D Cost | 802.1W (RSTP) Cost |
|---|---|---|
| 10 Mbps | 100 | 2,000,000 |
| 100 Mbps | 19 | 200,000 |
| 1 Gbps | 4 | 20,000 |
| 10 Gbps | 2 | 2,000 |
| 100 Gbps | 1 | 200 |
A DHCP scope defines the pool of addresses for a subnet. Reserve addresses for infrastructure (routers, switches, servers, printers) using exclusions. The DHCP scope size = Total usable hosts - Reserved static addresses. Ensure the pool supports peak concurrent clients plus 20% headroom.
SD-WAN link costing, cloud egress fees, NAT/PAT session capacity, and network downtime cost modeling.
Fast conversion tools for bits, bytes, speeds, and IP address formats — the most Googled networking conversions.
Network speeds are measured in bits per second (bps) while file sizes use bytes. The conversion factor is exactly 8 — but many confuse Megabits (Mb) with Megabytes (MB). A 100 Mbps internet connection transfers 100 ÷ 8 = 12.5 MB per second.
| Unit | Bits | Bytes | Common Use |
|---|---|---|---|
| 1 Kbps | 1,000 bits/s | 125 B/s | VoIP (G.729), dial-up |
| 1 Mbps | 1,000,000 bits/s | 125 KB/s | DSL, basic broadband |
| 1 Gbps | 1,000,000,000 bits/s | 125 MB/s | Gigabit LAN, fiber ISP |
| 1 Tbps | 10^12 bits/s | 125 GB/s | Backbone links, data centers |
| 1 MB file | 8,000,000 bits | 1,048,576 bytes | Takes 0.08s at 100 Mbps |
| 1 GB file | 8,000,000,000 bits | 1,073,741,824 bytes | Takes 80s at 100 Mbps |
Network uptime, SLA compliance, SNMP polling overhead, DNS TTL optimization, and downtime cost modeling.
Uptime % = (Total minutes - Downtime minutes) / Total minutes x 100
Annual downtime allowed = (1 - SLA%) x 525,600 minutes
99.9% SLA = 525.6 min/year = 43.8 min/month = 10.1 min/week
99.99% SLA = 52.56 min/year = 4.38 min/month = 1.01 min/week
99.999% SLA = 5.26 min/year = 26.3 sec/month = 6.05 sec/week
Fiber optic loss budgets, MPLS label stacks, link budgets, Fresnel zones, antenna gain, and PoE power planning.
A loss budget ensures your fiber link has enough optical power margin. Total loss must be less than the transceiver's power budget (transmit power minus receiver sensitivity).
Total Loss = (Fiber loss/km x Length) + (Connector loss x Connectors) + (Splice loss x Splices)
Typical values: SMF = 0.35 dB/km, MMF = 3 dB/km
Connector loss = 0.3-0.5 dB each, Fusion splice = 0.1 dB each
Power margin = Tx power - Rx sensitivity - Total loss (must be positive)
The Fresnel zone is an ellipsoid of space around the line-of-sight path. Obstructions in the first Fresnel zone cause significant signal diffraction and loss. You need 60% clearance of the first Fresnel zone radius for acceptable link performance.
r1 = 17.3 x sqrt(d1 x d2 / (f x (d1 + d2)))
where d1, d2 = distances from each end to obstacle (km), f = frequency (GHz)
Required clearance = 0.6 x r1 (60% of first Fresnel zone)
PoE switches have a total power budget shared across all ports. Plan for worst case: all ports simultaneously powered. PoE (802.3af) = 15.4W per port, PoE+ (802.3at) = 30W, PoE++ (802.3bt) = up to 90W. Total switch budget must exceed sum of all connected device wattage plus 20% headroom.
All formulas and reference data on this guide are sourced from current engineering standards and specifications: