GaussDB-AISQL: a composable cloud-native SQL system with AI capabilities

Abstract

Cloud-native data warehouses have revolutionized data analysis by enabling elasticity, high availability and lower costs. And the increasing popularity of artificial intelligence (AI) drives data warehouses to provide predictive analytics besides the existing descriptive analytics. Consequently, more vendors start to support training and inference of AI models in data warehouses, exploiting the benefits of near-data processing for fast model development and deployment. However, most of the existing solutions are limited by a complex syntax or slow data transportation across engines.

In this paper, we present GaussDB-AISQL, a composable SQL system with AI capabilities. GaussDB-AISQL adopts a composable system design that decouples computing, storage, caching, DB engine and AI engine. Our system offers all the functionality needed by end-to-end model training and inference during the model lifecycle. It also enjoys the simplicity and efficiency by providing a SQL-like syntax and removes the burden of manual model management. When training an AI model, GaussDB-AISQL benefits from highly parallel data transportation by concurrent data pulling from the distributed shared memory. The feature selection algorithms in GaussDB-AISQL make the training more data-efficient. When running model inference, GaussDB-AISQL registers the trained model object in the local data warehouse as a user-defined-function, which avoids moving inference data out of the data warehouse to an external AI engine. Experiments show that GaussDB-AISQL is up to 19× faster than baseline approaches.

Graphical abstract

Keywords

database system / data management / OLAP / cloud computing / AI / machine learning

Cheng Chen is now a PhD student at Renmin University of China, China. Currently he also works as an intern at the Database Innovation Lab of Huawei Cloud. His research interests are data-centric AI and DB for AI

Wenlong Ma is a research scientist at the Database Innovation Lab of Huawei Cloud. He received his PhD degree from Institute of Computing Technology, Chinese Academy of Sciences, China. His major research area lies in database systems and AI

Congli Gao is a research scientist at the Database Innovation Lab of Huawei Cloud, China. His major research area lies in database systems and AI

Wenliang Zhang is the director of the Database Innovation Lab of Huawei Cloud, China. His major research area lies in big data management systems and cloud computing

Kai Zeng is the Chief Architect of Huawei Cloud Data Warehouse Service. He also works as an adjunct professor in Yangtze Delta Region Institute, University of Electronic Science and Technology of China, China. His research interest lies in large scale data intensive systems

Tao Ye is a director at Huawei Cloud Data Warehouse Service. He holds a PhD in Computer Science from Huazhong University of Science and Technology, China. His research interests lie in exploring the fundamental principles and algorithms of database kernels

Yueguo Chen is a professor at School of Information, Renmin University of China, China. He received his PhD degree from National University of Singapore, Singapore. His research interests lie in database systems and interdisciplinary studies

Xiaoyong Du is a professor at School of Information, Renmin University of China, China. He is the director of the Key Laboratory of Data Engineering and Knowledge Engineering (Ministry of Education). His research interests lie in database systems, big data analytics, and knowledge engineering

References

[1] Marrandino, Alessandro. Machine Learning with BigQuery ML: Create, execute, and improve machine learning models in BigQuery using standard SQL queries. Packt Publishing Ltd, 2021

[2] Amazon Web Services, Inc. Amazon redshift machine learning. See docs.aws.amazoncom/redshift/latest/dg/machine_learning website, 2024

[3] Park K, Saur K, Banda D, Sen R, Interlandi M, Karanasos K. End-to-end optimization of machine learning prediction queries. In: Proceedings of 2022 International Conference on Management of Data, SIGMOD ’22. 2022, 587−601

[4] MindsDB. MindsDB. See mariadbcom/about-us/partners/mindsdb/ website, 2024

[5] Huang B, Babu S, Yang J. Cumulon: optimizing statistical data analysis in the cloud. In: Proceedings of 2013 ACM SIGMOD International Conference on Management of Data. 2013, 1−12

[6] Cohen J, Dolan B, Dunlap M, Hellerstein J M, Welton C . MAD skills: new analysis practices for big data. Proceedings of the VLDB Endowment, 2009, 2( 2): 1481–1492

[7] Lin Q, Wu S, Zhao J, Dai J, Li F, Chen G. A comparative study of in-database inference approaches. In: Proceedings of the 38th IEEE International Conference on Data Engineering (ICDE). 2022, 1794−1807

[8] Wang Y, Yang Y, Zhu W, Wu Y, Yan X, Liu Y, Wang Y, Xie L, Gao Z, Zhu W, Chen X, Yan W, Tang M, Tang Y. SQLFLow: a bridge between SQL and machine learning. 2020, arXiv preprint arXiv: 2001.06846

[9] Oracle Corporation. Oracle machine learning. See Docs.oracle.com/en/database/oracle/machine-learning/ website, 2024

[10] Wang D, Andres J, Weisz J D, Oduor E, Dugan C. AutoDS: towards human-centered automation of data science. In: Proceedings of 2021 CHI Conference on Human Factors in Computing Systems. 2021, 79

[11] Jordan M I, Mitchell T M . Machine learning: trends, perspectives, and prospects. Science, 2015, 349( 6245): 255–260

[12] Paganelli M, Sottovia P, Park K, Interlandi M, Guerra F . Pushing ML predictions into DBMSs. IEEE Transactions on Knowledge and Data Engineering, 2023, 35( 10): 10295–10308

[13] Substrait. See Github.com/substrait-io website, 2024

[14] Group T D M. The predictive model markup language. See dmg.org/pmml/pmml-v4-4-1.html website, 2024

[15] ONNX. See Onnx.ai/ website, 2024

[16] Chai C, Wang J, Tang N, Yuan Y, Liu J, Deng Y, Wang G. Efficient coreset selection with cluster-based methods. In: Proceedings of the 29th ACM SIGKDD Conference on Knowledge Discovery and Data Mining. 2023, 167−178

[17] Kumar A, Naughton J, Patel J M. Learning generalized linear models over normalized data. In: Proceedings of 2015 ACM SIGMOD International Conference on Management of Data. 2015, 1969−1984

[18] Kaggle. The state of data science. See www.kaggle.com/kaggle-survey-2020 website, 2020

[19] Psallidas F, Zhu Y, Karlas B, Interlandi M, Floratou A, Karanasos K, Wu W, Zhang C, Krishnan S, Curino C, Weimer M. Data science through the looking glass and what we found there. 2019, arXiv preprint arXiv: 1912.09536

[20] Grinsztajn L, Oyallon E, Varoquaux G. Why do tree-based models still outperform deep learning on typical tabular data? In: Proceedings of the 36th International Conference on Neural Information Processing Systems. 2022, 37

[21] The Apache Software Foundation. Apache arrow. See Arrow.apache website, 2016

[22] ClickHouse. ClickHouse. See githubcom/ClickHouse/ClickHouse website, 2024

[23] Apache Druid. Apache^® druid. See druidapache.org/ website, 2024

[24] MySQL. See www.mysql.com/ website, 2024

[25] Depoutovitch A, Chen C, Chen J, Larson P, Lin S, Ng J, Cui W, Liu Q, Huang W, Xiao Y, He Y. Taurus database: how to be fast, available, and frugal in the cloud. In: Proceedings of 2020 ACM SIGMOD International Conference on Management of Data. 2020, 1463−1478

[26] Ma Y, Xie S, Zhong H, Lee L, Lv K. HiEngine: how to architect a cloud-native memory-optimized database engine. In: Proceedings of 2022 International Conference on Management of Data. 2022, 2177−2190

[27] Shen J, Zuo P, Luo X, Su Y, Gu J, Feng H, Zhou Y, Lyu M R. Ditto: an elastic and adaptive memory-disaggregated caching system. In: Proceedings of the 29th Symposium on Operating Systems Principles. 2023, 675−691

[28] Achiam J, Adler S, Agarwal S, Ahmad L, Akkaya I, , . GPT-4 technical report. 2023, arXiv preprint arXiv: 2303.08774

[29] Ren X, Zhou P, Meng X, Huang X, Wang Y, Wang W, Li P, Zhang X, Podolskiy A, Arshinov G, Bout A, Piontkovskaya I, Wei J, Jiang X, Su T, Liu Q, Yao J. PanGu-Σ: Towards trillion parameter language model with sparse heterogeneous computing. 2023, arXiv preprint arXiv: 2303.10845

[30] Rojas J S. IP network traffic flows labeled with 75 apps. See Kaggle.com/datasets/jsrojas/ip-network-traffic-flows-labeled-with-87-apps website, 2018

[31] Kohavi R. Census income-UCI Machine Learning Repository. See Archive.ics.uci.edu/dataset/20/census+income website, 1996

[32] Bifet A, Ikonomovska E. The airlines dataset. See www.openml.org/d/1169 website, 2009

[33] Tromp J. Connect-4- UCI Machine Learning Repository. See Archive.ics.uci.edu/dataset/26/connect+4 website, 1995

[34] Moro S, Rita P, Cortez P. Bank marketing- UCI Machine Learning Repository. See Archive.ics.uci.edu/dataset/222/bank+marketing website, 2012

[35] Raabe M. The black Friday dataset. See www.openml.org website, 2019

[36] Mueller A. The diamonds dataset. See www.openml.org/data/download/21792853/dataset website, 2019

[37] Taxi N Y C. New York city taxi tip prediction. See www.openml.org/d/44065 website, 2016

[38] Group Mercedes Benz. Mercedes-Benz greener manufacturing. See Github.com/MezbanS/Mercedes-Benz-Greener-Manufacturing website, 2017

[39] Khamis M A, Ngo H Q, Nguyen X, Olteanu D, Schleich M . Learning models over relational data using sparse tensors and functional dependencies. ACM Transactions on Database Systems, 2020, 45( 2): 7

[40] Kadra A, Lindauer M, Hutter F, Grabocka J. Well-tuned simple nets excel on tabular datasets. In: Proceedings of the 35th International Conference on Neural Information Processing Systems. 2021, 1832

[41] Bej S, Davtyan N, Wolfien M, Nassar M, Wolkenhauer O . LoRAS: an oversampling approach for imbalanced datasets. Machine Learning, 2021, 110( 2): 279–301

[42] Kotelnikov A, Baranchuk D, Rubachev I, Babenko A. TabDDPM: modelling tabular data with diffusion models. In: Proceedings of the 40th International Conference on Machine Learning. 2023, 725

[43] Feurer M, Klein A, Eggensperger K, Springenberg J T, Blum M, Hutter F. Efficient and robust automated machine learning. In: Proceedings of the 28th International Conference on Neural Information Processing Systems. 2015, 2755−2763

[44] Yakovlev A, Moghadam H F, Moharrer A, Cai J, Chavoshi N, Varadarajan V, Agrawal S R, Idicula S, Karnagel T, Jinturkar S, Agarwal N . Oracle AutoML: a fast and predictive AutoML pipeline. Proceedings of the VLDB Endowment, 2020, 13( 12): 3166–3180

[45] Li Y, Shen Y, Zhang W, Zhang C, Cui B . VolcanoML: speeding up end-to-end AutoML via scalable search space decomposition. The VLDB Journal, 2023, 32( 2): 389–413

[46] H2O.ai. Scalable AutoML in H2O-3 open source. See H2o.ai/platform/h2o-automl/ website, 2023

[47] Patki N, Wedge R, Veeramachaneni K. The synthetic data vault. In: Proceedings of 2016 IEEE International Conference on Data Science and Advanced Analytics (DSAA). 2016, 399−410

[48] Pedreira P, Erling O, Karanasos K, Schneider S, McKinney W, Valluri S R, Zait M, Nadeau J . The composable data management system manifesto. Proceedings of the VLDB Endowment, 2023, 16( 10): 2679–2685

[49] Wilhite D. GoogleSQL: A SQL language as a component. In: Proceedings of the 1st International Workshop on Composable Data Management Systems. 2022

[50] Chattopadhyay B, Pedreira P, Agarwal S, Sun Y, Vakharia S, Li P, Liu W, Narayanan S. Shared foundations: modernizing Meta’s data lakehouse. In: Proceedings of the 13th Conference on Innovative Data Systems Research. 2023

[51] Begoli E, Camacho-Rodríguez J, Hyde J, Mior M J, Lemire D. Apache calcite: a foundational framework for optimized query processing over heterogeneous data sources. In: Proceedings of 2018 International Conference on Management of Data. 2018, 221−230

[52] Soliman M A, Antova L, Raghavan V, El-Helw A, Gu Z, Shen E, Caragea G C, Garcia-Alvarado C, Rahman F, Petropoulos M, Waas F, Narayanan S, Krikellas K, Baldwin R. Orca: a modular query optimizer architecture for big data. In: Proceedings of 2014 ACM SIGMOD International Conference on Management of Data. 2014, 337−348

[53] Pedreira P, Erling O, Basmanova M, Wilfong K, Sakka L, Pai K, He W, Chattopadhyay B . Velox: Meta’s unified execution engine. Proceedings of the VLDB Endowment, 2022, 15( 12): 3372–3384

[54] Microsoft. Microsoft SQL server machine learning services. website, 2024

[55] Karanasos K, Interlandi M, Psallidas F, Sen R, Park K, Popivanov I, Xin D, Nakandal S, Krishnan S, Weimer M, Yu Y, Ramakrishnan R, Curino C. Extending relational query processing with ML inference. In: Proceedings of the 10th Conference on Innovative Data Systems Research (CIDR 2020). 2020

[56] Corporation I. IBM db2 machine learning. website, 2024

[57] Li F . Modernization of databases in the cloud era: building databases that run like Legos. Proceedings of the VLDB Endowment, 2023, 16( 12): 4140–4151

[58] AP. SAP HANA predictive analysis library (PAL). See Help.sap.com website, 2024

[59] Hellerstein J M, Ré C, Schoppmann F, Wang D Z, Fratkin E, Gorajek A, Ng K S, Welton C, Feng X, Li K, Kumar A . The MADlib analytics library: or MAD skills, the SQL. Proceedings of the VLDB Endowment, 2012, 5( 12): 1700–1711

[60] Del Buono F, Paganelli M, Sottovia P, Interlandi M, Guerra F. Transforming ML predictive pipelines into SQL with MASQ. In: Proceedings of 2021 International Conference on Management of Data. 2021, 2696−2700

[61] Schule M, Lang H, Springer M, Kemper A, Neumann T, Gunnemann S. In-database machine learning with SQL on GPUs. In: Proceedings of the 33rd International Conference on Scientific and Statistical Database Management, SSDBM ’21. 2021, 25−36

[62] Olteanu D . The relational data Borg is learning. Proceedings of the VLDB Endowment, 2020, 13( 12): 3502–3515

[63] Gandhi A, Asada Y, Fu V, Gemawat A, Zhang L, Sen R, Curino C, Camacho-Rodríguez J, Interlandi M. The tensor data platform: towards an AI-centric database system. In: Proceedings of the 13th Conference on Innovative Data Systems Research. 2023

[64] Ghorbani M, Shaikhha A . Demonstration of OpenDBML, a framework for democratizing in-database machine learning. Proceedings of the VLDB Endowment, 2023, 16( 12): 3970–3973

[65] Miao H, Li A, Davis L S, Deshpande A. Towards unified data and lifecycle management for deep learning. In: Proceedings of the IEEE 33rd International Conference on Data Engineering (ICDE). 2017, 571−582

[66] Wang X, Dong X L, Meliou A. Data x-ray: a diagnostic tool for data errors. In: Proceedings of 2015 ACM SIGMOD International Conference on Management of Data. 2015, 1231−1245

[67] Vartak M, da Trindade J M F, Madden S, Zaharia M. MISTIQUE: a system to store and query model intermediates for model diagnosis. In: Proceedings of 2018 International Conference on Management of Data. 2018, 1285−1300

Acknowledgements

We thank the reviewers for their constructive feedback. This work was supported by the fund for building world-class universities (disciplines) of Renmin University of China.

Competing interests

The authors declare that they have no competing interests or financial conflicts to disclose.

RIGHTS & PERMISSIONS

2025 Higher Education Press