Speaker
Description
Current quantum computers (QC) are largely operated as single-user systems serving small research groups. Scaling access to thousands of users across the Europe, like institutions as part of the EuroHPC Joint Undertaking (EuroHPC JU), introduces two major challenges.
The first concerns multi-tenancy and resource governance. Users must be organized into projects with allocated quantum processing unit (QPU) time and policies governing resource consumption, while usage must be accurately accounted for. Although a QPU is inherently a serial resource, the infrastructure must support concurrent job submissions from multiple endpoints, efficient scheduling, and strict isolation of user data and results.
The second challenge is HPC–QC co-scheduling: coordinating workloads across classical HPC and quantum resources while accounting for their fundamentally different execution models and maximizing overall resource utilization.
To address these challenges and provide national and EuroHPC JU quantum computing resources to users, we developed QHEAppE, a Quantum-as-a-Service architecture integrating quantum resources with HEAppE Middleware.
The architecture combines four core components: the qaas Python client, HEAppE Middleware, QScheduler, and the LEXIS Platform. Together, they provide an end-to-end workflow covering user authentication and authorization, quantum workload submission and scheduling, QPU execution, result retrieval, and resource accounting.
Users submit quantum circuits through the QHEAppE client provided by the qaas Python package, supporting Qiskit, OpenQASM 3.0, and pulse-level playlists. Authentication and authorization are provided through the LEXIS Platform using MyAccessID, a federated identity provider proxy. HEAppE Middleware manages asynchronous job submission, lifecycle and state synchronization, and output retrieval, extending its HPC-as-a-Service abstraction to quantum systems.
HEAppE subsequently forwards jobs to QScheduler, a low-latency Rust-based scheduler forming the final scheduling layer between HEAppE and the QPU. QScheduler implements quantum-specific resource allocation and scheduling policies, including circuit grouping, job prioritization, resource reservations, runtime limits, and execution timeouts.
This architecture provides a scalable service layer for controlled, multi-user access to quantum computing resources and establishes a foundation for integrating quantum resources into distributed HPC environments.