Deploying a steered query optimizer in production at microsoft

W Zhang, M Interlandi, P Mineiro, S Qiao… - Proceedings of the …, 2022 - dl.acm.org
Modern analytical workloads are highly heterogeneous and massively complex, making
generic out of the box query optimizers untenable for many customers and scenarios. As a …

Optimizing the cloud? Don't train models. Build oracles!

T Bang, C Power, S Ameli, N Crooks… - arXiv preprint arXiv …, 2023 - arxiv.org
We propose cloud oracles, an alternative to machine learning for online optimization of
cloud configurations. Our cloud oracle approach guarantees complete accuracy and …

Conditionally risk-averse contextual bandits

M Farsang, P Mineiro, W Zhang - arXiv preprint arXiv:2210.13573, 2022 - arxiv.org
Contextual bandits with average-case statistical guarantees are inadequate in risk-averse
situations because they might trade off degraded worst-case behaviour for better average …

Dexter: A Performance-Cost Efficient Resource Allocation Manager for Serverless Data Analytics

AM Nestorov, D Marrón, A Gutierrez-Torre… - Proceedings of the 25th …, 2024 - dl.acm.org
Leveraging serverless platforms for the efficient execution of distributed data analytics
frameworks, such as Apache Spark [3], has gained substantial interest since early 2022. The …

The need for tabular representation learning: An industry perspective

J Cahoon, A Savelieva, AC Mueller… - NeurIPS 2022 First …, 2022 - openreview.net
The total addressable market for data applications has been estimated at\$70 B. This
includes the\$11 B market for data integration, which is estimated to grow at 25% in the …

Runtime Variation in Big Data Analytics

Y Zhu, R Sen, R Horton, JM Agosta - … of the ACM on Management of …, 2023 - dl.acm.org
The dynamic nature of resource allocation and runtime conditions on Cloud can result in
high variability in a job's runtime across multiple iterations, leading to a poor experience …

[PDF][PDF] Mitigating Tail Catastrophe in Steered Database Query Optimization with Risk-Averse Contextual Bandits

M Farsang, P Mineiro, W Zhang - mlforsystems.org
Contextual bandits with average-case statistical guarantees are inadequate in riskaverse
situations because they might trade off degraded worst-case behaviour for better average …