vDLT: A Service-Oriented Blockchain System with Virtualization and Decoupled Management/Control and Execution

F. Richard YuarXiv 2018 · arXiv

Abstract

Current blockchain systems often suffer from a "one-size-fits-all" approach that fails to accommodate diverse application requirements, leading to the well-known "Blockchain Trilemma." This paper proposes vDLT (virtualized Distributed Ledger Technology), a service-oriented architecture that leverages virtualization and the decoupling of control and execution layers. By hosting multiple virtual blockchains on a shared substrate, vDLT allows for tailored Quality of Service (QoS) levels, such as low latency or high throughput, optimized via Deep Reinforcement Learning (DRL).

Introduction: The Blockchain Trilemma and the Need for Differentiation

Mainstream blockchain platforms like Bitcoin and Ethereum struggle to balance decentralization, scalability, and security simultaneously. A primary cause is the lack of transaction differentiation; the network processes a simple payment with the same priority and resources as a complex industrial data log. This leads to congestion, high fees, and an inability to meet specific service-level agreements (SLAs).

The vDLT proposal argues that blockchain must evolve similarly to the early Internet—moving from a "best-effort" system to a sophisticated infrastructure capable of guaranteed service levels.

Proposed Methodology: A Layered and Virtualized Architecture

The core of vDLT is a multi-layered architecture inspired by Software-Defined Networking (SDN) and Network Function Virtualization (NFV).

1. Decoupled Management and Execution

To prevent the network's "brain" from being overwhelmed by processing tasks, vDLT separates the system into three distinct layers:

  • Application Layer: Where Decentralized Applications (DApps) reside.
  • Control Layer: Handles management tasks like prioritization, resource allocation, and node selection.
  • Execution Layer: Dedicated to running smart contracts and block production.
The three-layer architecture of vDLT, decoupling management/control from execution.

2. DLT Virtualization

Using a hypervisor-like approach, vDLT enables the dynamic creation of multiple virtual DLT systems (vDLTs) on top of a shared physical substrate. This allows "Virtual DLT Service Providers" (vDSPs) to offer specialized blockchain environments tailored to specific use cases.

Mapping of virtual DLT instances (vDLTs) to a shared substrate DLT via a virtualization layer.

3. Dynamic Resource Allocation and Consensus

  • Class of Service (CoS): Transactions are tagged with a CoS byte, categorizing them into eight service classes (e.g., 'fast confirmation', 'storage-intensive', 'low cost').
  • DRL-based Optimization: Deep Reinforcement Learning is employed to dynamically allocate CPU, storage, and bandwidth across the network to satisfy the QoS demands of various virtual chains.
  • Advanced Consensus: The system utilizes Delegated Proof-of-Stake (DPoS) with Quadratic Voting to ensure fair governance. Node-level consensus is achieved through an optimized Practical Byzantine Fault Tolerance (PBFT) protocol using EC-Schnorr multi-signatures to minimize communication overhead.

Key Insights

  • Service-Oriented Shift: Blockchain is reimagined as a flexible resource rather than a monolithic protocol. Class-Based Weighted Fair Queuing (CBWFQ) ensures that high-priority financial transactions are not delayed by lower-priority data.
  • Telecom Parallels: The evolution of vDLT mirrors the transition of cellular networks to 5G's Control/User Plane Separation, suggesting a necessary maturity path for decentralized infrastructure.
  • Hierarchical Scaling: A two-tier management hierarchy (local and global nodes) addresses the scalability challenges inherent in centralized management.

Limitations and Future Outlook

As a foundational design (Version 0.1), vDLT currently lacks implementation benchmarks. The overhead introduced by the virtualization layer and the real-world performance of the DRL-based allocator remain to be quantified. Future work will focus on developing a prototype to validate these architectural choices and conducting formal security analyses against common blockchain attack vectors.

Summary updated

BibTeX
@misc{yu2018vdlt,
  author = {F. Richard Yu},
  title = {{vDLT: A Service-Oriented Blockchain System with Virtualization and Decoupled Management/Control and Execution}},
  year = {2018},
  eprint = {1809.00290},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/1809.00290},
}
Figure