Skip to main content

Scaling

Introduction​

Liberdus inherits its scaling architecture from Shardus, which was built to overcome the limitations of traditional blockchain scaling. While most networks can only scale vertically or through limited forms of sharding, Liberdus combines multiple scaling mechanisms: horizontal scaling, linear scaling, and autoscaling, to ensure that it can grow efficiently, maintain low fees, and stay fully decentralized no matter how large the network becomes.

Each scaling layer serves a different purpose:

  • Horizontal Scaling increases raw capacity by adding more nodes.
  • Linear Scaling ensures that every new node proportionally increases throughput.
  • Autoscaling allows the network to grow or shrink dynamically based on real-time demand.

Together, these mechanisms make Liberdus one of the few decentralized systems capable of maintaining global scalability and affordability, even at massive transaction volumes.

Horizontal Scaling​

From a hardware perspective, Liberdus is designed to scale horizontally. Horizontal scaling is a method of increasing a network's capacity by adding more nodes. In contrast, vertical scalability increases capacity by upgrading existing nodes; adding extra RAM, CPU, or storage to the same machines. Liberdus uses horizontal scaling to increase Throughput (TPS), Storage and Bandwidth.

Each additional node added to the network directly increases the overall capacity.

Benefits of Horizontal Scaling​

  • Preserves decentralization, since new participants can join without needing expensive hardware.
  • Avoids hardware centralization, where only wealthy participants can afford to operate high-end nodes.
  • Enhances redundancy and reliability, improving fault tolerance.
  • Allows the network to handle sudden TPS surges by seamlessly incorporating additional nodes.

By scaling horizontally, Liberdus ensures sustainable growth while keeping node operation accessible, equitable, and censorship-resistant.

Linear Scaling​

Traditional Web3 networks that use network sharding, transaction sharding, or static state sharding have a fixed threshold where each shard becomes “full”. This threshold is determined by the maximum number of nodes permitted within a shard. Once the shard reaches that limit, a new shard must be created; a process known as stepwise scaling.

Limitations of Stepwise Scaling​

  • Requires manual intervention and planning.
  • May cause network disruptions or downtime.
  • Can introduce efficiency bottlenecks between shards.

In contrast, Liberdus employs linear scaling. With linear scaling, the network’s capacity increases proportionally with each new active node.

For example: If each active Liberdus node provides 2 TPS, and there are 1,000 active nodes, the total network capacity is 2,000 TPS. Adding two new nodes increases total throughput to 2,004 TPS, demonstrating perfectly predictable, linear growth.

Linear scaling allows Liberdus to maintain consistent performance, adapt dynamically, and expand without friction.

Autoscaling​

In major Web2 platforms, backend systems can dynamically scale up or down based on traffic. During high-demand periods (e.g., holiday sales), server capacity might increase fivefold; then automatically shrink when demand drops. This ensures performance remains stable while minimizing unnecessary operating costs.

However, most Web3 networks lack a decentralized version of autoscaling, leaving them inefficient and unable to adapt resource usage in real time.

Liberdus, however, inherits autoscaling directly from Shardus.

How Autoscaling Works in Liberdus​

  • The network automatically scales up (adding nodes) or down (removing nodes) to meet real-time resource needs.
  • A desired network size parameter determines optimal capacity.
  • Nodes independently and autonomously vote and reach consensus on whether the network should grow, shrink, or remain stable.
  • Adjustments are made according to current transaction load and storage requirements.

This means Liberdus can efficiently adapt to changing workloads, scaling globally during spikes in message volume and conserving energy during quieter times, without central coordination.