Data Stream Processing Systems Research Articles

SWIX: A Memory-efficient Sliding Window Learned Index

Data stream processing systems enable querying over sliding windows of streams of data. Efficient index structures for the streaming window are a crucial building block to enable querying the sliding window for operations such as aggregation and joins. This paper proposes SWIX, a novel memory-efficient learned index for sliding windows. Unlike conventional learned indexes that rely on tree structures to achieve logarithmic query cost, SWIX has a flat structure that uses substantially less memory and enables efficient query execution while having a low cost for index maintenance when inserting (and retraining). SWIX dynamically adapts itself to the real-time distribution shifts of data streams. SWIX outperforms existing indexes in terms of query execution time and memory footprint for workloads characterized by very frequent updates. Our results show that SWIX has a significantly smaller memory footprint than conventional, streaming, and learned indexes, using only 22% to 42% of the size compared to state-of-the-art approaches, yet outperforming them by up 1.2× to 1.6× on average (and up to 52×) in terms of query time, making it a space- and time-efficient method for indexing data streams. For concurrent learned indexes, Parallel SWIX can achieve up to 3.45× throughput with only 34% of memory consumption.

GT-scheduler: a hybrid graph-partitioning and tabu-search based task scheduler for distributed data stream processing systems

GT-scheduler: a hybrid graph-partitioning and tabu-search based task scheduler for distributed data stream processing systems

ContTune: Continuous Tuning by Conservative Bayesian Optimization for Distributed Stream Data Processing Systems

The past decade has seen rapid growth of distributed stream data processing systems. Under these systems, a stream application is realized as a Directed Acyclic Graph (DAG) of operators, where the level of parallelism of each operator has a substantial impact on its overall performance. However, finding optimal levels of parallelism remains challenging. Most existing methods are heavily coupled with the topological graph of operators, unable to efficiently tune under-provisioned jobs. They either insufficiently use previous tuning experience by treating successively tuning independently, or explore the configuration space aggressively, violating the Service Level Agreements (SLA). To address the above problems, we propose ContTune, a continuous tuning system for stream applications. It is equipped with a novel Big-small algorithm, in which the Big phase decouples the tuning from the topological graph by decomposing the job tuning problem into sub-problems that can be solved concurrently. We propose a conservative Bayesian Optimization (CBO) technique in the Small phase to speed up the tuning process by utilizing the previous observations. It leverages the state-of-the-art (SOTA) tuning method as conservative exploration to avoid SLA violations. Experimental results show that ContTune reduces up to 60.75% number of reconfigurations under synthetic workloads and up to 57.5% number of reconfigurations under real workloads, compared to the SOTA method DS2.

SMERAS -State Management with Efficient Resource Allocation and Scheduling in Big Data Stream Processing Systems

SMERAS -State Management with Efficient Resource Allocation and Scheduling in Big Data Stream Processing Systems

Dynamic Load Balancing in Stream Processing Pipelines Containing Stream-Static Joins

Data stream processing systems are used to continuously run mission-critical applications for real-time monitoring and alerting. These systems require high throughput and low latency to process incoming data streams in real time. However, changes in the distribution of incoming data streams over time can cause partition skew, which is defined as an unequal distribution of data partitions among workers, resulting in sub-optimal processing due to an unbalanced load. This paper presents the first solution designed specifically to address partition skew in the context of joining streaming and static data. Our solution uses state-of-the-art principles to monitor processing load, detect load imbalance, and dynamically redistribute partitions, to achieve optimal load balance. To accomplish this, our solution leverages the collocation of streaming and static data, while considering the processing load of the join and the subsequent stream processing operations. Finally, we present the results of an experimental evaluation, in which we compared the throughput and latency of four stream processing pipelines containing such a join. The results show that our solution achieved significantly higher throughput and lower latency than the competing approaches.

Virtual Commissioning of Distributed Systems in the Industrial Internet of Things

With the convergence of information technology (IT) and operational technology (OT) in Industry 4.0, edge computing is increasingly relevant in the context of the Industrial Internet of Things (IIoT). While the use of simulation is already the state of the art in almost every engineering discipline, e.g., dynamic systems, plant engineering, and logistics, it is less common for edge computing. This work discusses different use cases concerning edge computing in IIoT that can profit from the use of OT simulation methods. In addition to enabling machine learning, the focus of this work is on the virtual commissioning of data stream processing systems. To evaluate the proposed approach, an exemplary application of the middleware layer, i.e., a multi-agent reinforcement learning system for intelligent edge resource allocation, is combined with a physical simulation model of an industrial plant. It confirms the feasibility of the proposed use of simulation for virtual commissioning of an industrial edge computing system using Hardware-in-the-Loop. In summary, edge computing in IIoT is highlighted as a new application area for existing simulation methods from the OT perspective. The benefits in IIoT are exemplified by various use cases for the logic or middleware layer using physical simulation of the target environment. The relevance for real-life IIoT systems is confirmed by an experimental evaluation, and limitations are pointed out.

Dynamic Performance Analysis of STEP System in Internet of Vehicles Based on Queuing Theory.

The Internet of vehicles (IoV) is an important research area of the intelligent transportation systems using Internet of things theory. The complex event processing technology is a basic issue for processing the data stream in IoV. In recent years, many researchers process the temporal and spatial data flow by complex event processing technology. Spatial Temporal Event Processing (STEP) is a complex event query language focusing on the temporal and spatial data flow in Internet of vehicles. There are four processing models of the event stream processing system based on the complex event query language: finite automata model, matching tree model, directed acyclic graph model, and Petri net model. In addition, the worst-case response time of the event stream processing system is an important indicator of evaluating the performance of the system. Firstly, this paper proposed a core algorithm of the temporal and spatial event stream processing program based on STEP by Petri net model. Secondly, we proposed a novel method to estimate the worst-case response time of the event stream processing system, which is based on stochastic Petri net and queuing theory. Finally, through the simulation experiment based on queuing theory, this paper proves that the data stream processing system based on STEP has good dynamic performance in processing the spatiotemporal data stream in Internet of vehicles.

Runtime Adaptation of Data Stream Processing Systems: The State of the Art

Data stream processing (DSP) has emerged over the years as the reference paradigm for the analysis of continuous and fast information flows, which often have to be processed with low-latency requirements to extract insights and knowledge from raw data. Dealing with unbounded dataflows, DSP applications are typically long running and thus, likely experience varying workloads and working conditions over time. To keep a consistent service level in face of such variability, a lot of effort has been spent studying strategies for runtime adaptation of DSP systems and applications. In this survey, we review the most relevant approaches from the literature, presenting a taxonomy to characterize the state of the art along several key dimensions. Our analysis allows us to identify current research trends as well as open challenges that will motivate further investigations in this field.

Resource Configuration Tuning for Stream Data Processing Systems via Bayesian Optimization

Stream data processing systems are becoming increasingly popular in the big data era. Systems such as Apache Flink typically provide a number (e.g., 30) of configuration parameters to flexibly specify the amount of resources (e.g., CPU cores and memory) allocated for tasks. These parameters significantly affect task performance. However, it is hard to manually tune them for optimal performance for an unknown program running on a given cluster. An automatic as well as fast resource configuration tuning approach is therefore desired. To this end, we propose to leverage Bayesian optimization to automatically tune the resource configurations for stream data processing systems. We first select a machine learning model—Random Forest—to construct accurate performance models for a stream data processing program. We subsequently take the Bayesian optimization (BO) algorithm, along with the performance models, to iteratively search the optimal configurations for a stream data processing program. Experimental results show that our approach improves the 99th-percentile tail latency by a factor of 2.62× on average and up to 5.26× overall. Furthermore, our approach improves throughput by a factor of 1.05× on average and up to 1.21× overall.

QoS-Aware Approximate Query Processing for Smart Cities Spatial Data Streams.

Large amounts of georeferenced data streams arrive daily to stream processing systems. This is attributable to the overabundance of affordable IoT devices. In addition, interested practitioners desire to exploit Internet of Things (IoT) data streams for strategic decision-making purposes. However, mobility data are highly skewed and their arrival rates fluctuate. This nature poses an extra challenge on data stream processing systems, which are required in order to achieve pre-specified latency and accuracy goals. In this paper, we propose ApproxSSPS, which is a system for approximate processing of geo-referenced mobility data, at scale with quality of service guarantees. We focus on stateful aggregations (e.g., means, counts) and top-N queries. ApproxSSPS features a controller that interactively learns the latency statistics and calculates proper sampling rates to meet latency or/and accuracy targets. An overarching trait of ApproxSSPS is its ability to strike a plausible balance between latency and accuracy targets. We evaluate ApproxSSPS on Apache Spark Structured Streaming with real mobility data. We also compared ApproxSSPS against a state-of-the-art online adaptive processing system. Our extensive experiments prove that ApproxSSPS can fulfill latency and accuracy targets with varying sets of parameter configurations and load intensities (i.e., transient peaks in data loads versus slow arriving streams). Moreover, our results show that ApproxSSPS outperforms the baseline counterpart by significant magnitudes. In short, ApproxSSPS is a novel spatial data stream processing system that can deliver real accurate results in a timely manner, by dynamically specifying the limits on data samples.

I-Scheduler: Iterative scheduling for distributed stream processing systems

Task allocation in Data Stream Processing Systems (DSPSs) has a significant impact on performance metrics such as data processing latency and system throughput. An application processed by DSPSs can be represented as a Directed Acyclic Graph (DAG), where each vertex represents a task and the edges show the dataflow between the tasks. Task allocation can be defined as the assignment of the vertices in the DAG to the physical compute nodes such that the data movement between the nodes is minimised. Finding an optimal task placement for DSPSs is NP-hard. Thus, approximate scheduling approaches are required to improve the performance of DSPSs. In this paper, we propose a heuristic scheduling algorithm which reliably and efficiently finds highly communicating tasks by exploiting graph partitioning algorithms and a mathematical optimisation software package. We evaluate the communication cost of our method using three micro-benchmarks, showing that we can achieve results that are close to optimal. We further compare our scheduler with two popular existing schedulers, R-Storm and Aniello et al.’s ‘Online scheduler’ using two real-world applications. Our experimental results show that our proposed scheduler outperforms R-Storm, increasing throughput by up to 30%, and improves on the Online scheduler by 20%–86% as a result of finding a more efficient schedule.11This work is an extension of I-Scheduler (Eskandari et al., 2018 [1]).

POTUS: Predictive Online Tuple Scheduling for Data Stream Processing Systems

Most online service providers deploy their own data stream processing systems in the cloud to conduct large-scale and real-time data analytics. However, such systems, e.g., Apache Heron, often adopt naive scheduling schemes to distribute data streams (in the units of tuples) among processing instances, which may result in workload imbalance and system disruption. Hence, there still exists a mismatch between the temporal variations of data streams and such inflexible scheduling scheme designs. Besides, the fundamental limits of benefits of predictive scheduling to data stream processing systems remain unexplored. In this article, we focus on the problem of tuple scheduling with predictive service in Apache Heron. With a careful choice in the granularity of system modeling and decision making, we formulate the problem as a stochastic network optimization problem and propose <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">POTUS</i> , an online predictive scheduling scheme that aims to minimize the response time of data stream processing by steering data streams in a distributed fashion. Theoretical analysis and simulation results show that POTUS achieves an ultra-low response time with a stability guarantee. Moreover, POTUS only requires mild-value of future information to effectively reduce the response time, even with mis-prediction.

Flink-ER: An Elastic Resource-Scheduling Strategy for Processing Fluctuating Mobile Stream Data on Flink

As real-time and immediate feedback becomes increasingly important in tasks related to mobile information, big data stream processing systems are increasingly applied to process massive amounts of mobile data. However, when processing a drastically fluctuating mobile data stream, the lack of an elastic resource-scheduling strategy limits the elasticity and scalability of data stream processing systems. To address this problem, this paper builds a flow-network model, a resource allocation model, and a data redistribution model as the foundation for proposing Flink with an elastic resource-scheduling strategy (Flink-ER), which consists of a capacity detection algorithm, an elastic resource reallocation algorithm, and a data redistribution algorithm. The strategy improves the performance of the platform by dynamically rescaling the cluster and increasing the parallelism of operators based on the processing load. The experimental results show that the throughput of a cluster was promoted under the premise of meeting latency constraints, which verifies the efficiency of the strategy.

Evaluating the impact of a coordinated checkpointing in distributed data streams processing systems using discrete event simulation

Data Streams Processing systems process continuous flows of data under Quality of Service requirements. Data streams often contain critical information which requires real-time processing. To guarantee systems' dependability and avoid information loss, one must use a fault-tolerance strategy. However, there are several strategies available, and the proper evaluation of which mechanism is better for each system architecture is challenging, especially in large-scale distributed systems. In this paper, we propose a discrete simulation model for investigating the impacts of the Coordinated Checkpoint fault tolerance strategy imposes on Data Stream Processing Systems. Results show that this strategy critically affects stream processing in failure-prone situations due to an increase in latency up to 120% and information loss, reaching 95% of the processing window in the worst case.

A Survey on Automatic Parameter Tuning for Big Data Processing Systems

Big data processing systems (e.g., Hadoop, Spark, Storm) contain a vast number of configuration parameters controlling parallelism, I/O behavior, memory settings, and compression. Improper parameter settings can cause significant performance degradation and stability issues. However, regular users and even expert administrators grapple with understanding and tuning them to achieve good performance. We investigate existing approaches on parameter tuning for both batch and stream data processing systems and classify them into six categories: rule-based, cost modeling, simulation-based, experiment-driven, machine learning, and adaptive tuning. We summarize the pros and cons of each approach and raise some open research problems for automatic parameter tuning.

Hardware-Conscious Stream Processing

Data stream processing systems (DSPSs) enable users to express and run stream applications to continuously process data streams. To achieve realtime data analytics, recent researches keep focusing on optimizing the system latency and throughput. Witnessing the recent great achievements in the computer architecture community, researchers and practitioners have investigated the potential of adoption hardware-conscious stream processing by better utilizing modern hardware capacity in DSPSs. In this paper, we conduct a systematic survey of recent work in the field, particularly along with the following three directions: 1) computation optimization, 2) stream I/O optimization, and 3) query deployment. Finally, we advise on potential future research directions.

L-Heron: An open-source load-aware online scheduler for Apache Heron

Abstract Apache Heron has emerged as a promising Data Stream Processing System (DSPS). However, it lacks intelligent scheduling strategy, which results in significant performance degradation for streaming applications in certain scenarios. In this paper, we first illustrate the inefficiencies and challenges of Heron default scheduling in current practice through experimental observations and analysis. Motivated by our observations, we propose L -Heron, an online scheduler based on Heron, which has the following features: (i) based on runtime information, it can improve the data processing efficiency by using the load-aware online scheduling, which heuristically minimizes the overall communication overhead by identifying the traffic load; (ii) it is load aware, which can effectively balance the workload of a topology to avoid heavy performance loss caused by overloading of worker nodes; (iii) it provides an online scheduling interface that is transparent to users, which allows users to focus on their scheduling logic and easily deploy them to the system. Additionally, we have evaluated L-Heron on well-known example topologies and a realistic application. Extensive experimental results show that the effectiveness of L-Heron is consistent among multiple metrics including the system completion latency, inter-node traffic, CPU utilization and throughput, with respect to Heron and recent related work.

DSPBench: A Suite of Benchmark Applications for Distributed Data Stream Processing Systems

Systems enabling the continuous processing of large data streams have recently attracted the attention of the scientific community and industrial stakeholders. Data Stream Processing Systems (DSPSs) are complex and powerful frameworks able to ease the development of streaming applications in distributed computing environments like clusters and clouds. Several systems of this kind have been released and currently maintained as open source projects, like Apache Storm and Spark Streaming. Some benchmark applications have often been used by the scientific community to test and evaluate new techniques to improve the performance and usability of DSPSs. However, the existing benchmark suites lack of representative workloads coming from the wide set of application domains that can leverage the benefits offered by the stream processing paradigm in terms of near real-time performance. The goal of this article is to present a new benchmark suite composed of 15 applications coming from areas like Finance, Telecommunications, Sensor Networks, Social Networks and others. This article describes in detail the nature of these applications, their full workload characterization in terms of selectivity, processing cost, input size and overall memory occupation. In addition, it exemplifies the usefulness of our benchmark suite to compare real DSPSs by selecting Apache Storm and Spark Streaming for this analysis.

Challenges and Solutions for Processing Real-Time Big Data Stream: A Systematic Literature Review

Contribution: Recently, real-time data warehousing (DWH) and big data streaming have become ubiquitous due to the fact that a number of business organizations are gearing up to gain competitive advantage. The capability of organizing big data in efficient manner to reach a business decision empowers data warehousing in terms of real-time stream processing. A systematic literature review for real-time stream processing systems is presented in this paper which rigorously look at the recent developments and challenges of real-time stream processing systems and can serve as a guide for the implementation of real-time stream processing framework for all shapes of data streams. Background: Published surveys and reviews either cover papers focusing on stream analysis in applications other than real-time DWH or focusing on extraction, transformation, loading (ETL) challenges for traditional DWH. This systematic review attempts to answer four specific research questions. Research Questions: 1)Which are the relevant publication channels for real-time stream processing research? 2) Which challenges have been faced during implementation of real-time stream processing? 3) Which approaches/tools have been reported to address challenges introduced at ETL stage while processing real-time stream for real-time DWH? 4) What evidence have been reported while addressing different challenges for processing real-time stream? Methodology: A systematic literature was conducted to compile studies related to publication channels targeting real-time stream processing/joins challenges and developments. Following a formal protocol, semi-automatic and manual searches were performed for work from 2011 to 2020 excluding research in traditional data warehousing. Of 679,547 papers selected for data extraction, 74 were retained after quality assessment. Findings: This systematic literature highlights implementation challenges along with developed approaches for real-time DWH and big data stream processing systems and provides their comparisons. This study found that there exists various algorithms for implementing real-time join processing at ETL stage for structured data whereas less work for un-structured data is found in this subject matter.

Stream Data Load Prediction for Resource Scaling Using Online Support Vector Regression

A distributed data stream processing system handles real-time, changeable and sudden streaming data load. Its elastic resource allocation has become a fundamental and challenging problem with a fixed strategy that will result in waste of resources or a reduction in QoS (quality of service). Spark Streaming as an emerging system has been developed to process real time stream data analytics by using micro-batch approach. In this paper, first, we propose an improved SVR (support vector regression) based stream data load prediction scheme. Then, we design a spark-based maximum sustainable throughput of time window (MSTW) performance model to find the optimized number of virtual machines. Finally, we present a resource scaling algorithm TWRES (time window resource elasticity scaling algorithm) with MSTW constraint and streaming data load prediction. The evaluation results show that TWRES could improve resource utilization and mitigate SLA (service level agreement) violation.