Calculate how many network packets per second your connection can handle based on speed and packet size (MTU).
Start ConvertingCalculate how many network packets per second your connection can handle based on speed and packet size (MTU).
Watch the conversion happen in real-time as you adjust the speed slider.
Understanding packets, frames, PPS, and how they differ from Mbps.
A packet is a unit of data transmitted over a network. Every email, web page, and video stream is broken into packets before being sent. Each packet contains a header (source/destination address, protocol info) and a payload (the actual data). Typical packet sizes range from 64 bytes to 1,500 bytes on standard Ethernet networks.
A frame is a Layer 2 (Data Link) container that wraps a packet for Ethernet transmission. It adds an Ethernet header (14 bytes), an optional VLAN tag (4 bytes), and a CRC trailer (4 bytes). The minimum Ethernet frame is 64 bytes; maximum standard frame is 1,518 bytes (or 9,022 bytes for jumbo frames).
PPS (Packets Per Second) measures how many discrete network packets a device or link can process each second. Unlike Mbps which measures raw bit throughput, PPS reflects the processing overhead per packet — every packet requires header parsing, routing table lookups, ACL checks, and forwarding decisions regardless of its size.
Mbps measures the volume of bits flowing per second — like water volume through a pipe. PPS measures individual packets processed — like counting water bottles. A 1 Gbps link carries 83,333 PPS at 1,500B or 1.95 million PPS at 64B. Same bandwidth, vastly different PPS demands on your hardware.
Calculate Network Packets Per Second. Here's the formula and a step-by-step example.
Calculate how many network packets per second your connection can handle based on speed and packet size (MTU).
PPS = (Mbps × 1,000,000) ÷ (Packet Size × 8) — The conversion factor is Speed ÷ Packet.
PPS is critical across networking — from home routers to cloud data centers and DDoS defense.
Every router has a maximum PPS rating. A home router handles 10,000–15,000 PPS for web browsing, but a data center core router must process 100+ million PPS. Exceeding PPS limits causes packet drops and severe congestion.
Firewalls inspect every packet header and payload. A firewall rated at 1 Gbps throughput might only handle 50,000–100,000 PPS with deep packet inspection enabled. Small-packet floods overwhelm firewalls long before bandwidth limits are reached.
Online games send small packets (64–256 bytes) at high frequency (20–128 tick rate). A competitive FPS server with 64 players at 128-tick sends 8,192 packets/second. PPS bottlenecks cause lag spikes and rubber-banding.
Voice calls use G.711 codec sending 50 packets/second per call. An office PBX with 200 concurrent calls generates 10,000 PPS of small voice packets. Insufficient PPS capacity causes jitter, echo, and dropped calls.
AWS, Azure, and GCP instances have PPS limits. An AWS EC2 m5.large supports ~100,000 PPS while m5.24xlarge handles 4+ million PPS. Wrong instance size throttles app performance regardless of bandwidth.
Load balancers process every incoming packet for routing decisions. A DDoS attack using 64-byte packets at 100 Gbps generates 148 million PPS — overwhelming devices even if bandwidth seems manageable. PPS capacity determines DDoS resilience.
Quick reference chart for common Mbps to PPS conversions.
Understanding the critical differences between bandwidth, throughput, latency, packet loss, and PPS.
Bandwidth is the maximum theoretical data rate of a link, measured in Mbps or Gbps. Think of it as the width of a highway — a 1 Gbps link is a 1,000-lane highway for bits. Bandwidth tells you the capacity, not the actual speed of individual data units.
Throughput is the actual data rate achieved after accounting for overhead, retransmissions, and protocol inefficiency. A 1 Gbps link typically delivers 940–960 Mbps of real throughput. Throughput is always lower than bandwidth due to headers, flow control, and congestion.
Latency is the time delay for a single packet to travel from source to destination, measured in milliseconds (ms). Low latency (under 20 ms) is critical for gaming and VoIP. High latency doesn't reduce PPS but increases round-trip time, affecting interactive applications.
Packet loss occurs when packets are dropped before reaching their destination. Common causes: buffer overflow when PPS exceeds device capacity, congestion, faulty hardware, or wireless interference. Even 1% packet loss can reduce TCP throughput by up to 50%.
Most speed tests show bandwidth and latency but ignore PPS. A device might pass a 1 Gbps throughput test with large packets but choke on 100,000 small packets per second. PPS is the bottleneck metric for VoIP, gaming, IoT, microservices, and DDoS mitigation. Always evaluate PPS alongside bandwidth when sizing network equipment.
Common packet sizes and their real-world uses in networking.
The smallest valid Ethernet frame. Used by ARP requests, ICMP pings, TCP ACKs, and keepalive messages. Generates the highest PPS for any given bandwidth. At 1 Gbps: 1,953,125 PPS. Critical stress test size for DDoS and firewall benchmarks.
Common for VoIP voice packets (G.711 codec) and game state updates. Contains a small payload (~66 bytes of data after headers). At 1 Gbps: 976,562 PPS. Frequent in real-time applications that prioritize latency over bulk transfer.
Typical for DNS queries and responses, SNMP traps, and small API payloads. At 1 Gbps: 244,140 PPS. A balanced size between overhead efficiency and real-time responsiveness.
Common in web browsing, email, and general TCP traffic. Many HTTP response chunks fall in this range. At 1 Gbps: 122,070 PPS. A practical "average" for mixed enterprise networks.
The default Maximum Transmission Unit for Ethernet. Used by most file transfers, streaming, downloads, and web traffic. At 1 Gbps: 83,333 PPS. The standard baseline for PPS calculations and the most common packet size on the internet.
Used in data center SAN/NAS traffic, iSCSI, and server-to-server transfers. Reduces PPS by 6× compared to 1500B, lowering CPU overhead. At 1 Gbps: 13,888 PPS. Requires end-to-end jumbo frame support on all switches and NICs.
Every Ethernet frame carries protocol overhead that reduces effective payload throughput and affects PPS calculations.
Real-world PPS calculations for common network scenarios.
Suitable for: Home routers, small office firewalls, VPN gateways, WiFi access points. Most consumer-grade hardware can handle 10,000–30,000 PPS easily.
Suitable for: Enterprise next-gen firewalls, DDoS scrubbing appliances, carrier-grade switches, 10GbE network monitoring taps. Requires ASIC-based forwarding or DPDK-accelerated software.
How PPS affects real networking scenarios across industries.
Why PPS matters more than bandwidth for understanding and defending against DDoS attacks.
DDoS attackers deliberately use minimum-size 64-byte packets because each packet requires the same CPU processing as a full 1,500-byte packet — header parsing, ACL checks, connection tracking, and forwarding decisions.
A firewall rated at 10 Gbps throughput but only 1 million PPS will be overwhelmed by a 5 Gbps flood of 64-byte packets (generating 9.7 million PPS). The PPS limit is hit at 50% of bandwidth capacity.
Modern DDoS mitigation requires hardware-accelerated packet processing (ASIC/FPGA) capable of 100+ Mpps, not just high-bandwidth pipes.
How different network hardware handles packet processing — from NICs to CPUs.
Routers forward packets between networks using routing tables. Software routers (pfSense, VyOS) handle 100K–1M PPS on modern CPUs. Hardware routers (Cisco, Juniper) use ASICs for 100M–1B+ PPS at line rate with zero CPU involvement.
Firewalls inspect packet headers and payloads. Stateless filtering: 1M+ PPS on mid-range hardware. Stateful inspection: drops to 200K–500K PPS. Deep Packet Inspection (DPI): may drop to 50K–100K PPS. Each layer of inspection halves PPS capacity.
1 GbE NICs: ~1.49 Mpps (64B line-rate). 10 GbE NICs: ~14.88 Mpps. 25/40 GbE NICs: 37–59 Mpps. Modern NICs use RSS (Receive Side Scaling) to distribute packets across CPU cores and hardware offloading for checksum and segmentation.
A single CPU core using the Linux kernel stack processes ~200K–500K PPS. With DPDK (Data Plane Development Kit) kernel bypass: 5M–15M PPS per core. With XDP (eXpress Data Path) in eBPF: 10M+ PPS per core. CPU clock speed and L3 cache size directly impact PPS.
Layer 2/3 switches use ASIC-based forwarding for wire-speed processing. A 48-port 1 GbE switch handles ~71.4 Mpps (all ports at line-rate with 64B frames). A 32-port 100 GbE data center switch processes 4.76 billion PPS. Switches are the highest-PPS devices in any network.
Understanding Maximum Transmission Unit sizes, jumbo frames, and packet fragmentation.
When a packet exceeds the MTU of a link, it must be fragmented into smaller pieces. Each fragment becomes its own packet with its own header, increasing PPS and overhead.
Don't Fragment (DF) bit to prevent silent fragmentation.How a packet flows through network infrastructure from source to destination.
Each hop in the chain has its own PPS limit. The lowest PPS device becomes the bottleneck for the entire path.
PPS capabilities of popular router and firewall platforms.
Avoid these frequent errors when calculating or evaluating PPS.
Mbps and PPS are fundamentally different metrics. A 1 Gbps link can mean 83,333 PPS (1,500B packets) or 1,953,125 PPS (64B packets). Comparing devices by Mbps alone ignores their packet processing limits.
Calculating PPS without specifying packet/frame size is meaningless. Always state the MTU or packet size alongside PPS numbers. "100K PPS" could mean 100K at 1,500B (efficient) or 100K at 64B (stressed).
Forgetting IP (20B), TCP (20B), or UDP (8B) headers inflates payload estimates. A 1,500-byte Ethernet frame carries only ~1,460 bytes of TCP payload. A 64-byte frame has only ~18 bytes of usable payload.
Wire-level overhead adds 38 bytes per frame (preamble 7B, SFD 1B, CRC 4B, inter-frame gap 12B, Ethernet header 14B). This means the true on-wire size of a "1,500-byte" packet is actually 1,538 bytes.
When calculating PPS for hardware sizing, always use the total frame size on the wire — including all headers and overhead — not just the application payload. Using payload size overestimates PPS capacity by 3–40% depending on packet size. For accurate calculations, add all Layer 2, 3, and 4 headers to your payload size.
Understanding the relationship between Internet Speed & Packet Size and Packets per Second.
Converting Mbps to PPS helps you understand your actual data throughput. ISPs advertise in Mbps but your experience depends on PPS.
Many applications and protocols specify bandwidth in PPS. Use this converter to match your network capacity to software requirements.
PPS = (Mbps × 1,000,000) ÷ (Packet Size × 8). Apply Speed ÷ Packet to any Mbps value. For example: 100 Mbps = 8,333 PPS (1500B) PPS.
Memorize the factor: Speed ÷ Packet. This lets you do instant conversions in your head whenever you see Mbps values.
Common questions about converting Mbps to PPS.
PPS (Packets Per Second) measures how many network packets a link can carry each second. It depends on both bandwidth and packet size.
Ethernet MTU is 1,500 bytes. Minimum Ethernet frame is 64 bytes. Jumbo frames can be 9,000 bytes.
Routers and firewalls have PPS limits. Small packets (VoIP, gaming) stress PPS more than large packets (file transfers).
Several factors reduce real-world PPS: Ethernet overhead (preamble, inter-frame gap, CRC) adds ~38 bytes per frame, protocol headers (IP + TCP/UDP) consume payload space, and hardware limitations in your NIC, router, or firewall cap the processing rate.
It depends on the use case. Home routers handle 8,000–15,000 PPS. Enterprise firewalls handle 1–10 Mpps. Data center switches process 100+ Mpps. For VoIP, you need at least 50 PPS per call; for gaming servers, 64–128 PPS per player.
At 1,500-byte MTU: 1 Gbps = 83,333 PPS. At 64-byte minimum frames: 1 Gbps = 1,953,125 PPS (~1.95 Mpps). The smaller the packet, the higher the PPS.
Firewalls have a maximum PPS rating independent of bandwidth. A firewall rated at 500 Mbps might only handle 50,000 PPS. Small-packet floods (like DDoS with 64-byte packets) hit the PPS limit before the bandwidth limit, dropping legitimate packets.
Games send many small, frequent packets (64–256 bytes at 20–128 Hz). High PPS demand means if your router hits its PPS limit, packets queue up causing lag spikes and rubber-banding.
VoIP codecs send a 160-byte packet every 20 ms (50 PPS/call). With 100 calls = 5,000 PPS. If your network can't process packets fast enough, some arrive late (jitter) or drop, causing choppy audio and dropped calls.