Skip to main content
ExplainerNetwork ArchitectureExplainer· 4 min read· in Gaming & Esports

How Delta Compression Reduces Bandwidth Requirements for 100-Player Battle Royale Servers

By transmitting only the changes in game state rather than full snapshots, delta compression allows massive multiplayer servers to synchronize 100 players at 60Hz without collapsing under gigabit bandwidth loads.

By Jackson Reed

Network Architects 40%Esports Competitors 30%Cloud Infrastructure Providers 30%
Network Architects
Engineers focused on optimizing the trade-off between server CPU load and outbound bandwidth.
Esports Competitors
Players who experience the downstream effects of compression through latency and desync.
Cloud Infrastructure Providers
Hosting services that manage the physical hardware and egress costs of massive multiplayer games.

Perspectives this story doesn't cover

  • Telecommunication ISPs managing the aggregate load of millions of concurrent gamers.
  • Engineers working on deterministic lockstep models, an alternative to state synchronization.

A 100-player server ticking at 60 hertz generates 6,000 individual state updates every single second. If a game engine transmits the full positional data, rotation, and velocity for every player in the lobby—roughly 100 bytes per entity—the server must push 600 kilobytes per second to a single client. Scaled across the entire lobby, that single match demands 60 megabytes per second, or 480 megabits per second, of continuous outbound bandwidth. At standard cloud hosting egress rates, running a single battle royale match would be financially ruinous, and the sheer volume of data would saturate the network interfaces of standard server blades before the game even began.[2]

The mechanism that prevents this mathematical collapse is delta compression. Rather than broadcasting a complete snapshot of the world 60 times a second, the server transmits only the data that has changed since the last update the client acknowledged receiving. If a player is standing still, their positional coordinates are omitted entirely from the next packet. If they are moving in a predictable straight line, the server might only send the velocity vector once, allowing the client to extrapolate the rest of the movement locally until an input changes.[1]

This approach fundamentally alters the economics and technical feasibility of massive multiplayer environments. According to documentation from the Valve Developer Community, early multiplayer architectures were constrained by dial-up modems that could handle no more than 5 to 7 kilobytes per second. By implementing delta compression—a technique popularized by John Carmack during the development of Quake 3—developers reduced the payload of a standard update from hundreds of bytes to a fraction of that size, allowing complex 3D environments to synchronize over heavily constrained networks.[1]

Transmitting only changed state data reduces server-side bandwidth requirements by over 95 percent.

The compression pipeline begins with quantization. A standard floating-point number, which dictates a player's exact coordinate in a 3D space, occupies 32 bits of data. As outlined by networking engineer Glenn Fiedler, sending that full 32-bit float for every minor movement is highly inefficient. Instead, engines quantize the data, stripping away sub-millimeter precision that the human eye cannot perceive on a monitor. A coordinate like 123.456789 is truncated, significantly reducing the baseline size of the number before the actual delta comparison even begins.[2]

A standard floating-point number, which dictates a player's exact coordinate in a 3D space, occupies 32 bits of data.

Once the data is quantized, the engine applies Huffman encoding to the resulting deltas. Huffman coding is a variable-length compression algorithm that assigns the shortest possible bit sequences to the most common values. In a competitive shooter, the most common change in a player's vertical velocity over a 16-millisecond tick is zero. Under Huffman encoding, that zero can be transmitted as a single bit. If 80 players in a 100-player lobby are not jumping, the server spends exactly 80 bits—just 10 bytes—updating their vertical state for the entire server.[3]

To make this work, the server maintains a rolling history of acknowledged states for every connected client. When a client receives packet 104, it sends an acknowledgement back to the server. The server then calculates the delta between packet 104 and the current state at packet 108, transmitting only the difference. If a packet is dropped due to network jitter, the server simply calculates the next delta against the last known good state, ensuring the client eventually receives the correct geometry without forcing a full, bandwidth-heavy resynchronization.[1]

After an initial full state snapshot, subsequent delta packets remain extremely small unless a player experiences severe packet loss.

This architecture is not without significant hardware trade-offs. Delta compression requires the server to hold the historical state of every client in memory, drastically increasing the CPU and RAM overhead on the host machine. A Source engine server running at a 100Hz tick rate generates roughly 50 percent more CPU load than one running at 66Hz, largely due to the computational cost of comparing thousands of historical states against the current frame to generate custom delta packets for each individual player.[1]

Despite the computational cost, the bandwidth savings are non-negotiable for modern esports. Without delta compression, the 480 Mbps requirement for a 100-player lobby would cause cascading packet loss and unplayable latency for anyone without a commercial fiber connection. By reducing the per-client bandwidth requirement to roughly 15 to 30 kilobytes per second, delta compression allows publishers to host dozens of concurrent matches on a single machine, making the free-to-play battle royale model physically and financially viable across the globe.[4]

Key points

  • A 100-player server sending uncompressed data at 60Hz would require roughly 480 Mbps of outbound bandwidth.
  • Delta compression solves this by transmitting only the data that has changed since the last acknowledged packet.
  • Techniques like quantization and Huffman encoding further reduce the size of the transmitted data.
  • The bandwidth savings come at the cost of significantly higher CPU and RAM usage on the server.

Why this matters

Without delta compression, the bandwidth required to host a 100-player match would make massive multiplayer games financially impossible to operate. By reducing data payloads by over 90%, this networking technique forms the invisible backbone of the modern free-to-play esports industry.

Key terms

Delta Compression
A networking technique that transmits only the changes in data since the last update, rather than sending the entire state every time.
Quantization
The process of reducing the precision of a number, such as rounding a coordinate, to make it smaller and easier to compress.
Huffman Encoding
A compression algorithm that assigns the shortest bit sequences to the most frequently occurring values, minimizing data size.
Tick Rate
The frequency at which a game server simulates the world and broadcasts updates to clients, usually measured in hertz (Hz).

Frequently asked

Why do games still lag if delta compression is so efficient?

Delta compression reduces bandwidth, but it cannot fix packet loss or high ping. If the physical connection drops packets, the client must wait for the next successful delta update, causing visual stutter.

Does a higher tick rate increase bandwidth?

Yes. A 128Hz server sends twice as many updates as a 64Hz server. While delta compression keeps each packet small, the sheer volume of packets still increases the total bandwidth and CPU load.

How does the server know what the client has received?

Clients send acknowledgement messages (ACKs) back to the server for every packet. The server uses the last acknowledged packet as the baseline to calculate the next delta.

Sources

Source coverage

4 outlets

3 viewpoints surfaced

Network Architects 40%Esports Competitors 30%Cloud Infrastructure Providers 30%
  1. [1]Valve Developer CommunityNetwork Architects

    Source Multiplayer Networking

    Read on Valve Developer Community
  2. [2]Gaffer on GamesNetwork Architects

    State Synchronization

    Read on Gaffer on Games
  3. [3]WikipediaCloud Infrastructure Providers

    Data compression

    Read on Wikipedia
  4. [4]Factlen Editorial TeamEsports Competitors

    Synthesis by Factlen editorial team

    Read on Factlen Editorial Team

Comments

Stay informed

Every angle. Every day.

Get Gaming & Esports stories with full source coverage and perspective breakdowns delivered to your inbox.