Graph Neural Networks (GNNs) are widely used for learning on graph-structured data, but scaling GNN training to massive graphs remains challenging. To enable scalable distributed training, graphs are divided into smaller partitions that are distributed across multiple machines such that inter-machine communication is minimized and computational load is balanced. In practice, existing partitioning approaches face a fundamental trade-off between partitioning overhead and partitioning quality. We propose EmbedPart, an embedding-driven partitioning approach that achieves both speed and quality. Instead of operating directly on irregular graph structures, EmbedPart leverages node embeddings produced during the actual GNN training workload and clusters these dense embeddings to derive a partitioning. EmbedPart achieves more than 100x speedup over Metis while maintaining competitive partitioning quality and accelerating distributed GNN training. Moreover, EmbedPart naturally supports graph updates and fast repartitioning, and can be applied to graph reordering to improve data locality and accelerate single-machine GNN training. By shifting partitioning from irregular graph structures to dense embeddings, EmbedPart enables scalable and high-quality graph data optimization.

Existing Stream Processing Engines (SPEs), designed for scale-out in the cloud, often overlook the need for scaleup performance on heterogeneous hardware. As a result, they provide latency and throughput guarantees at the expense of high overhead on system resources, for example, memory and CPU usage. This over-provisioning is tolerable in cloud environments where resources are generously available. However, it is detrimental for low-end devices that are becoming increasingly common in today's cloud-edge environments. In this paper, we present the architecture of our adaptive and efficient multi-query stream processing engine, which is the core of the open-source system NebulaStream. Our engine is optimized for resource-constrained devices and their special requirements. In particular, our engine achieves better hardware utilization due to several key engine design decisions: 1) a task-based execution model for multi-query, scale-up execution based on an order-aware data flow model, 2) work-stealing-aware operators using hardware-tailored code generation, 3) runtime compute caching, and 4) work-stealing-aware state management. As a result, our novel engine outperforms general-purpose systems by at least one order of magnitude, executes hundreds of queries in parallel without performance degradation, and seamlessly reacts to fluctuating ingestion rates.

Foundation Models (FMs) are increasingly integrated into remote sensing (RS) pipelines. These models include unimodal vision encoders and multimodal architectures. FMs are adapted to diverse perception tasks, such as image classification, change detection, and visual question answering. However, selecting the most suitable remote sensing foundation model (RSFM) for a specific task remains challenging due to scattered documentation, heterogeneous formats, and complex deployment constraints. To address this, we first introduce the RSFM Database (RS-FMD), the first structured and schema-guided resource covering over 160 RSFMs trained on various data modalities, spanning different spatial, spectral, and temporal resolutions, considering different learning paradigms. Built upon RS-FMD, we further present REMSA, a constraint-aware agent that enables automated RSFM selection from natural language queries. REMSA combines structured FM metadata retrieval with a task-driven decision workflow. In detail, it interprets user input, clarifies missing constraints, ranks models via in-context learning, and provides transparent justifications. Our system supports various RS tasks and data modalities, enabling personalized, reproducible, and efficient FM selection. To evaluate REMSA, we construct a benchmark of 100 expert-verified RS query scenarios. Each query is evaluated across 4 systems and 3 LLM backbones, with the top-3 selected models manually assessed by domain experts. This results in 3,000 expert-scored task–system–model configurations under our novel expert-centered evaluation protocol. REMSA outperforms multiple baselines, showing its practical utility in real decision-making applications. REMSA operates entirely on publicly available metadata of open source RSFMs, without accessing private or sensitive data.

Real-time data processing in large geo-distributed applications, like the Internet of Things (IoT), increasingly shifts computation from the cloud to the network edge to reduce latency and mitigate network congestion. In this setting, minimizing latency while avoiding node overload requires jointly optimizing operator replication and placement of operator instances, a challenge known as the Operator Placement and Replication (OPR) problem. OPR is NP-hard and particularly difficult to solve in large-scale, heterogeneous, and dynamic geo-distributed networks, where solutions must be scalable, resource-aware, and adaptive to changes like node failures. Existing work on OPR has primarily focused on single-stream operators, such as filters and aggregations. However, many latency-sensitive applications, like environmental monitoring and anomaly detection, require efficient regional stream joins near data sources. This paper introduces Nova, an optimization approach designed to address OPR for join operators that are computable on resource-constrained edge devices. Nova relaxes the NP-hard OPR into a convex optimization problem by embedding cost metrics into a Euclidean space and partitioning joins into smaller sub-joins. This new formulation enables linear scalability and efficient adaptation to topological changes through partial re-optimizations. We evaluate Nova through simulations on real-world topologies and on a local testbed, demonstrating up to 39x latency reduction and 4.5x increase in throughput compared to existing edge-centered solutions, while also preventing node overload and maintaining near-constant re-optimization times regardless of topology size.

This paper presents refined BigEarthNet (reBEN) that is a large-scale, multi-modal remote sensing dataset constructed to support deep learning (DL) studies for remote sensing image analysis. The reBEN dataset consists of 549,488 pairs of Sentinel-1 and Sentinel-2 image patches. To construct reBEN, we initially consider the Sentinel-1 and Sentinel-2 tiles used to construct the BigEarthNet dataset and then divide them into patches of size 1200m x 1200 m. We apply atmospheric correction to the Sentinel-2 patches using the latest version of the sen2cor tool, resulting in higher-quality patches compared to those present in BigEarthNet. Each patch is then associated with a pixel-level reference map and scene-level multi-labels. This makes reBEN suitable for pixel- and scene-based learning tasks. The labels are derived from the most recent CORINE Land Cover (CLC) map of 2018 by utilizing the 19-class nomenclature as in BigEarthNet. The use of the most recent CLC map results in overcoming the label noise present in BigEarthNet. Furthermore, we introduce a new geographical-based split assignment algorithm that significantly reduces the spatial correlation among the train, validation, and test sets with respect to those present in BigEarthNet. This increases the reliability of the evaluation of DL models. To minimize the DL model training time, we introduce a software tool (called as rico-hdl) that converts the reBEN dataset into a DL-optimized data format. In our experiments, we show the potential of reBEN for multi-modal multi-label image classification problems by considering several state-of-theart DL models. The pre-trained model weights, associated code, and complete dataset are available at https://bigearth.net.

Spatial data is ubiquitous. Massive amounts of data are generated every day from a plethora of sources such as billions of GPS-enabled devices (e.g., cell phones, cars, and sensors), consumer-based applications (e.g., Uber and Strava), and social media platforms (e.g., location-tagged posts on Facebook, Twitter, and Instagram). This exponential growth in spatial data has led the research community to build systems and applications for efficient spatial data processing. In this study, we apply a recently developed machine-learned search technique for single-dimensional sorted data to spatial indexing. Specifically, we partition spatial data using six traditional spatial partitioning techniques and employ machine-learned search within each partition to support point, range, distance, and spatial join queries. Adhering to the latest research trends, we tune the partitioning techniques to be instance-optimized. By tuning each partitioning technique for optimal performance, we demonstrate that: (i) grid-based index structures outperform tree-based index structures (from 1.23$\times$ to 2.47$\times$), (ii) learning-enhanced variants of commonly used spatial index structures outperform their original counterparts (from 1.44$\times$ to 53.34$\times$ faster), (iii) machine-learned search within a partition is faster than binary search by 11.79% - 39.51% when filtering on one dimension, (iv) the benefit of machine-learned search diminishes in the presence of other compute-intensive operations (e.g. scan costs in higher selectivity queries, Haversine distance computation, and point-in-polygon tests), and (v) index lookup is the bottleneck for tree-based structures, which could potentially be reduced by linearizing the indexed partitions.

On October 19 and 20, 2023, the authors of this report convened in Cambridge, MA, to discuss the state of the database research field, its recent accomplishments and ongoing challenges, and future directions for research and community engagement. This gathering continues a long standing tradition in the database community, dating back to the late 1980s, in which researchers meet roughly every five years to produce a forward looking report. This report summarizes the key takeaways from our discussions. We begin with a retrospective on the academic, open source, and commercial successes of the community over the past five years. We then turn to future opportunities, with a focus on core data systems, particularly in the context of cloud computing and emerging hardware, as well as on the growing impact of data science, data governance, and generative AI. This document is not intended as an exhaustive survey of all technical challenges or industry innovations in the field. Rather, it reflects the perspectives of senior community members on the most pressing challenges and promising opportunities ahead.

The Internet of Things (IoT) demands real-time, low-latency processing of data generated by thousands of heterogeneous, resourceconstrained devices. In such dynamic environments, ensuring fault tolerance becomes critical, especially for safety-sensitive applications like disaster management or patient monitoring. However, challenges across large, hierarchically connected IoT topologies. In this paper, we present Meerkat, a network-aware fault-tolerance protocol explicitly designed for IoT environments. Meerkat achieves zero-downtime recovery via redundant operator execution on disjoint paths and efficient duplicate detection. It also includes dynamic load balancing that adapts operator placement to device volatility, ensuring fair resource use. Compared to state-of-the-art techniques, Meerkat sustains up to 70 & times; higher throughput with only 28% network overhead. These results highlight Meerkat's ability to deliver efficient fault tolerance with minimal overhead at IoT scale.

NebulaStream is a novel, open-source data stream processing system for massively distributed, heterogeneous data streams in the cloud-edge continuum. It adheres to the design goals of ease-of-use, extensibility, and efficiency to provide a framework for users and developers to implement diverse Internet of Things (IoT) use cases. Equipped with essential built-in functionalities, NebulaStream allows users to customize the system easily while ensuring efficient execution even on low-end devices. In this demonstration, we highlight NebulaStream's ability to integrate and process multi-modal, multi-frequency data streams. We showcase its abilities through a real-world IoT scenario where NebulaStream is used to improve the health assessment of patients in a smart intensive care unit.

Data management systems research focuses on improving human and technical efficiency for performing data analysis tasks. In this presentation, I describe selected research contributions to achieve that goal. I first highlight work on using query feedback in order to improve the cardinality model of a relational query optimizer. I then will discuss how the research vision of the Stratosphere research project at TU Berlin lead to the creation of the data stream processing system Apache Flink. As a third contribution, I will discuss how fractal space-filling curves can be used to efficiently process multidimensional range queries. I will conclude by giving an outlook on NebulaStream, a novel data processing system to handle massively distributed data streams on heterogeneous devices.