SAM: Subseries Augmentation-based Meta-learning for Generalizing AIOps Model in Multi-Cloud Migration

Abstract

In the context of cloud computing, enterprises are increasingly adopting multi-cloud strategies to enhance performance, ensure cost efficiency, and avoid vendor lock-in. This trend poses a significant challenge for the migration of AI for IT operations (AIOps) models across different cloud providers due to variations in architecture, performance, and data distribution. Traditional methods of re-training AIOps models on new cloud environments are labor-intensive and delay deployment. To address this issue, we introduce a novel framework, i.e., SAM (Subseries Augmentation-based Meta-learning), aiming to facilitate seamless model migration between clouds without the need for re-training from scratch. SAM utilizes data augmentation and meta-learning to adapt AIOps models to new cloud environments efficiently. In this paper, we demonstrate SAM's effectiveness in adapting anomaly detectors across different configurations on both public and simulated datasets. We assert that SAM can be adapted to other AI models that are used to automate IT tasks such as alerting and resource scaling.