An Elastic Multi-Instance DID Architecture via Layer 2 Scaling for Web3.0 Identity Services

Abstract

The widespread adoption of decentralized identity (DID) is constrained by blockchain scalability issues. Mainstream Layer 1 blockchains are costly, while generic Layer 2 Rollups face the “Noisy Neighbor” problem, where identity operations compete with DeFi traffic, causing unpredictable delays. To address this, we propose EM-DID, an Elastic Multiinstance DID Architecture. Unlike static chains, EM-DID treats Layer 2 execution environments as dynamic resources. The system employs three novel mechanisms: an Auto-Scaling Manager utilizing a Warm Pool strategy for near-zero coldstart latency and Hysteresis to prevent resource oscillation; a QoS-Aware Scheduling Algorithm that routes requests based on cost-latency trade-offs; and a Hierarchical State Aggregation Protocol that compresses proofs from multiple instances into a single Layer 1 transaction. Experiments demonstrate that EM-DID achieves linear scalability. While a single simulation instance sustains 600 TPS constrained by the testbed environment, the architecture supports horizontal scaling to thousands of TPS by adding instances without degradation. The architecture reduces gas costs by 99.76% compared to Layer 1 and minimizes average latency by approximately 94% through intelligent routing. EM-DID provides a viable path for deploying high-frequency, self-sovereign identity services in scenarios such as the Internet of Things (IoT).