OPTIMUM-DERAM: Highly Consistent, Scalable, and Secure Multi-Object Memory using RLNC

TL;DR

OPTIMUM-DERAM addresses the scalability and security challenges of distributed atomic memory by combining RLNC-based erasure coding with a consistent hashing ring for multi-object placement and a blockchain-based oracle for dynamic participation. It builds a Byzantine-tolerant MWMR protocol and extends it to dynamic, multi-object settings with join/depart mechanisms that migrate data without sacrificing liveness. The approach demonstrates improved latency, storage efficiency, and resilience compared to traditional ABD-like solutions, and scales across object size, object count, node count, and concurrent operations. This provides a practical, decentralized memory service suitable for Web3 and distributed applications where nodes are geographically dispersed and subject to churn and adversarial behavior.

Abstract

This paper introduces OPTIMUM-DERAM, a highly consistent, scalable, secure, and decentralized shared memory solution. Traditional distributed shared memory implementations offer multi-object support by multi-threading a single object memory instance over the same set of data hosts. While theoretically sound, the amount of resources required made such solutions prohibitively expensive in practical systems. OPTIMUM-DERAM proposes a decentralized, reconfigurable, atomic read/write shared memory (DeRAM) that: (i) achieves improved performance and storage scalability by leveraging Random Linear Network Codes (RLNC); (ii) scales in the number of supported atomic objects by introducing a new object placement and discovery approach based on a consistent hashing ring; (iii) scales in the number of participants by allowing dynamic joins and departures leveraging a blockchain oracle to serve as a registry service; and (iv) is secure against malicious behavior by tolerating Byzantine failures. Experimental results over a globally distributed set of nodes, help us realize the performance and scalability gains of OPTIMUM-DERAM over previous distributed shared memory solutions (i.e., the ABD algorithm [3])

Figures (5)

• Figure 1: Original data is encoded by multiplying with a random matrix, here $A$ of dimension $8 \times 8$.
• Figure 2: The flexnodes and objects are logically placed in a ring. Here for the example we are assuming $N=2^5$.
• Figure 3: 52 Optimum-DeRAM nodes used in numerical evaluations
• Figure 4: Read (solid) and write (dashed) execution latency. (Left) Varying object sizes. (Middle) Varying number of objects stored in the DSM. (Right) Varying number of nodes. In all instances, 10 readers and 3 writers executed requests concurrently in an interval of $10s$
• Figure 5: Read and write latencies under varying concurrency in a DSM with 13 nodes and 100 objects. (Left) Varying number of readers (writers fixed at 10). (Middle) Varying number of writers (readers fixed at 10). Mean memory consumption for varying number of objects stored in the DSM. (Right)

Theorems & Definitions (41)

• Lemma 3.1: Fault Tolerance
• proof
• Lemma 3.2: Concurrency Factor
• proof
• Theorem 3.3: Liveness
• proof
• Definition 3.4: DAP Consistency Properties
• Lemma 3.5: Property DAP1
• proof
• Lemma 3.6: Property DAP2