DB4ML - An In-Memory Database Kernel with Machine Learning Support

Jasny, Matthias; Ziegler, Tobias; Kraska, Tim; Roehm, Uwe; Binnig, Carsten

doi:10.1145/3318464.3380575

Cited by 25 publications

(11 citation statements)

References 22 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Another relevant issue is investigating a learned strategy [33][34][35] to implement the adaptive mechanism in AMG-Buffer. It has been a hot topic in recent years to use machine learning models for optimizing database components [36]. In future work, it will be possible to consider a machine-learning-based approach to provide a more general solution to selecting the migration granularity for page replacements.…”

Section: Discussionmentioning

confidence: 99%

Adaptive Multi-Grained Buffer Management for Database Systems

Wang

Jin

2021

Future Internet

View full text Add to dashboard Cite

The traditional page-grained buffer manager in database systems has a low hit ratio when only a few tuples within a page are frequently accessed. To handle this issue, this paper proposes a new buffering scheme called the AMG-Buffer (Adaptive Multi-Grained Buffer). AMG-Buffer proposes to use two page buffers and a tuple buffer to organize the whole buffer. In this way, the AMG-Buffer can hold more hot tuples than a single page-grained buffer. Further, we notice that the tuple buffer may cause additional read I/Os when writing dirty tuples into disks. Thus, we introduce a new metric named clustering rate to quantify the hot-tuple rate in a page. The use of the tuple buffer is determined by the clustering rate, allowing the AMG-Buffer to adapt to different workloads. We conduct experiments on various workloads to compare the AMG-Buffer with several existing schemes, including LRU, LIRS, CFLRU, CFDC, and MG-Buffer. The results show that AMG-Buffer can significantly improve the hit ratio and reduce I/Os compared to its competitors. Moreover, the AMG-Buffer achieves the best performance on a dynamic workload as well as on a large data set, suggesting its adaptivity and scalability to changing workloads.

show abstract

Section: Discussionmentioning

confidence: 99%

Adaptive Multi-Grained Buffer Management for Database Systems

Wang

Jin

2021

Future Internet

View full text Add to dashboard Cite

show abstract

“…Later in the 2010's, MADlib [27] suggested using UDAs/UDFs for embedding ML into the database. Apache Spark's MLlib [47], Apache Mahout Samsara [75] and others [23,30,40,77,82,83,88] can be seen as a continuation of this effort. A broad overview of in-DB ML techniques can be found in [39].…”

Section: Related Workmentioning

confidence: 99%

End-to-end Optimization of Machine Learning Prediction Queries

Park

Saur

Banda

et al. 2022

Proceedings of the 2022 International Conference on Management of Data

View full text Add to dashboard Cite

Prediction queries are widely used across industries to perform advanced analytics and draw insights from data. They include a data processing part (e.g., for joining, filtering, cleaning, featurizing the datasets) and a machine learning (ML) part invoking one or more trained models to perform predictions. These parts have so far been optimized in isolation, leaving significant opportunities for optimization unexplored.We present Raven, a production-ready system for optimizing prediction queries. Raven follows the enterprise architectural trend of collocating data and ML runtimes. It relies on a unified intermediate representation that captures both data and ML operators in a single graph structure to unlock two families of optimizations. First, it employs logical optimizations that pass information between the data part (and the properties of the underlying data) and the ML part to optimize each other. Second, it introduces logical-tophysical transformations that allow operators to be executed on different runtimes (relational, ML, and DNN) and hardware (CPU, GPU). Novel data-driven optimizations determine the runtime to be used for each part of the query to achieve optimal performance. Our evaluation shows that Raven improves performance of prediction queries on Apache Spark and SQL Server by up to 13.1× and 330×, respectively. For complex models where GPU acceleration is beneficial, Raven provides up to 8× speedup compared to state-of-the-art systems.

show abstract

“…BAGUA is built on decades of research regarding distributed machine learning systems and algorithms. Plenty of them are from the database community [52,53,54,55,56,57,58,59,60,61,46,47]. We now summarize related work and discuss some in details to provide backgrounds and contexts.…”

Section: Preliminaries and Related Workmentioning

confidence: 99%

BAGUA: Scaling up Distributed Learning with System Relaxations

Gan¹,

Lian²,

Wang³

et al. 2021

Preprint

View full text Add to dashboard Cite

Recent years have witnessed a growing list of systems for distributed data-parallel training. Existing systems largely fit into two paradigms, i.e., parameter server and MPI-style collective operations. On the algorithmic side, researchers have proposed a wide range of techniques to lower the communication via "system relaxations": quantization, decentralization, and communication delay. However, most, if not all, existing systems only rely on standard synchronous and asynchronous stochastic gradient (SG) based optimization, therefore, cannot take advantage of all possible optimizations that the machine learning community has been developing recently. Given this emerging gap between the current landscapes of systems and theory, we build BAGUA, a communication framework whose design goal is to provide a system abstraction that is both flexible and modular to support state-of-the-art system relaxation techniques of distributed training. Powered by the new system design, BAGUA has a great ability to implement and extend various state-ofthe-art distributed learning algorithms. In a production cluster with up to 16 machines (128 GPUs), BAGUA can outperform PyTorch-DDP, Horovod and BytePS in the end-to-end training time by a significant margin (up to 1.95×) across a diverse range of tasks. Moreover, we conduct a rigorous tradeoff exploration showing that different algorithms and system relaxations achieve the best performance over different network conditions.

show abstract

DB4ML - An In-Memory Database Kernel with Machine Learning Support

Cited by 25 publications

References 22 publications

Adaptive Multi-Grained Buffer Management for Database Systems

Adaptive Multi-Grained Buffer Management for Database Systems

End-to-end Optimization of Machine Learning Prediction Queries

BAGUA: Scaling up Distributed Learning with System Relaxations

Contact Info

Product

Resources

About