Distributed Systems Architecture

How
Optimum
Works

A coded peer-to-peer network built on random linear network coding, decentralized shared memory, and threshold-based reconstruction — designed for maximum resilience.

Explore the System Read Concepts
Optimum network visualization
k/n Threshold Logic
∞ Encoded Fragments
0 Single Points of Failure

Step by Step

Follow the journey of data through the Optimum network — from creation to reconstruction.

1 / 6

Core
Concepts

Three interlocking primitives that make Optimum fundamentally different from traditional P2P networks. Each concept builds on the last to create a system that is more than the sum of its parts.

01 — Encoding
MSG α₁s α₂s α₃s
RLNC
Random Linear Network Coding transforms a message into algebraically combined fragments. Each shard is a random linear combination of the source data — interchangeable, redundant, and independently useful.
"Data becomes interchangeable fragments — any k of n reconstructs the whole."
02 — Transport
mumP2P
Coded peer-to-peer communication where nodes never send raw messages — only encoded combinations. Every transmission contributes to the network's knowledge without revealing the original data structure.
"Nodes send encoded data instead of full messages — always useful, never raw."
03 — Memory
DeRAM
Decentralized shared memory that abstracts the network into a unified address space. Nodes read and write without knowing the physical location of data — it behaves like RAM, distributed across peers.
"Acts like shared memory across the network — transparent, consistent, decentralized."

Parameter
Tuning

Three levers control the behavior of the entire system. Adjust them to trade off between redundancy, latency, and network density.

01 — Fragmentation
Shards (n)
8 total shards
Number of encoded fragments generated per message. Higher n increases redundancy but raises bandwidth usage per node.
02 — Recovery
Threshold (k)
5 of n required
Minimum shards required to reconstruct the original message. Lower k = higher fault tolerance. k/n ratio defines resilience.
03 — Topology
Mesh Density
6 peer connections
Number of peers each node maintains connections with. Denser meshes propagate faster but consume more resources.
System State
Balanced
Redundancy 62%
Fault Tolerance 37%
Propagation Speed 50%
Current configuration provides a balanced trade-off between redundancy and efficiency.

Real-World
Behavior

Observed performance characteristics under simulated network conditions with packet loss, node churn, and variable latency.

Reconstruction Latency vs. Packet Loss Rate

Simulation: 8 shards, threshold k=5, 12-node mesh — 500ms window

Reconstruction Success Rate
94.2%
Avg. Latency (ms)
14ms
Bandwidth Efficiency
78%
Node Fault Tolerance
3 of 8 nodes