Logo image
Back
Policy-Aware Adaptive Data Processing Framework for Multi-Tenant Cloud Analytics Platforms
Conference paper

Policy-Aware Adaptive Data Processing Framework for Multi-Tenant Cloud Analytics Platforms

Jitendra Gopaluni, Voolla Sandeep Kumar, Satvik Bhasin and Pushpanjali Chauhan
Innovations in Machine, Engineering, and Digital Conference (IMED), pp 1-5
06 Mar 2026
DOI: https://doi.org/10.1109/IMED68921.2026.11484415

Abstract

Adaptive Data Processing Central Processing Unit Circuits Cloud analytics Communication systems Electronic circuits Feedback Frequency modulation Internet Multi Tenant Neuromorphics PA-ADPF Policy-aware Radio broadcasting Internet of Things
Multi-tenant cloud analytics systems have to be able to handle heterogeneous workloads effectively and observe tenant-specific policies, including service-level agreements, data locality, security and cost limits. Old methods of managing resources are not suitable to address the dynamic load and the policy requirements. This paper presents a Policy-Aware Adaptive Data Processing Framework (PA-ADPF), which combines machine learning algorithms to collaboratively achieve performance and policy fidelity in multi-tenant cloud systems. The framework applies supervised learning to predict the risk of policy violation, regression to approximate the number of resources needed, and reinforcement learning to allow adaptive decisions regarding scheduling. Empirical analysis of simulated multi-tenant workloads indicates that the proposed framework is highly effective to reduce the execution latency and policy violations and to enhance the throughput and resource utilization in comparison with the baseline approaches. The findings support the idea that integrating policy awareness into machine learning (ML)based decision-making allows scalable and efficient with average latency as 295 ms, throughput is 690 jobs per hour and resource utilization is 83%.

Metrics

1 Record Views

Details

Logo image