More Like this

1. Exploring the Potential of Next Generation Software-Defined In-Memory Frameworks

   https://doi.org/10.1109/SBAC-PAD.2018.00042  

   Chen, Shouwei ; Rodero, Ivan ( January 2018 , 2018 30th International Symposium on Computer Architecture and High Performance Computing)

   As in-memory data analytics become increasingly important in a wide range of domains, the ability to develop large-scale and sustainable platforms faces significant challenges related to storage latency and memory size constraints. These challenges can be resolved by adopting new and effective formulations and novel architectures such as software-defined infrastructure. This paper investigates the key issue of data persistency for in-memory processing systems by evaluating persistence methods using different storage and memory devices for Apache Spark and the use of Alluxio. It also proposes and evaluates via simulation a Spark execution model for using disaggregated off- rack memory and non-volatile memory targeting next-generation software-defined infrastructure. Experimental results provide better understanding of behaviors and requirements for improving data persistence in current in-memory systems and provide data points to better understand requirements and design choices for next-generation software-defined infrastructure. The findings indicate that in-memory processing systems can benefit from ongoing software-defined infrastructure implementations; however current frameworks need to be enhanced appropriately to run efficiently at scale. 

   more » « less
2. OpenMSIStream: A Python package for facilitatingintegration of streaming data in diverse laboratory environments

   https://doi.org/10.21105/joss.04896  

   Eminizer, Margaret ; Tabrisky, Sam ; Sharifzadeh, Amir ; DiMarco, Christopher ; Diamond, Jacob M. ; Ramesh, K. T. ; Hufnagel, Todd C. ; McQueen, Tyrel M. ; Elbert, David ( March 2023 , Journal of Open Source Software)

   OpenMSIStream provides seamless connection of scientific data stores with streaming infrastructure to allow researchers to leverage the power of decoupled, real-time data streaming architectures. Data streaming is the process of transmitting, ingesting, and processing data continuously rather than in batches. Access to streaming data has revolutionized many industries in the past decade and created entirely new standards of practice and types of analytics. While not yet commonly used in scientific research, data streaming has the potential to become a key technology to drive rapid advances in scientific data collection (e.g., Brookhaven National Lab (2022)). This paucity of streaming infrastructures linking complex scientific systems is due to a lack of tools that facilitate streaming in the diverse and distributed systems common in modern research. OpenMSIStream closes this gap between underlying streaming systems and common scientific infrastructure. Closing this gap empowers novel streaming applications for scientific data including automation of data curation, reduction, and analysis; real-time experiment monitoring and control; and flexible deployment of AI/ML to guide autonomous research. Streaming data generally refers to data continuously generated from multiple sources and passed in small packets (termed messages). Streaming data messages are typically organized in groups called topics and persist for periods of time conducive to processing for multiple uses either sequentially or in small groups. The resulting flows of raw data, metadata, and processing results form “ecosystems” that automate varied data-driven tasks. A strength of data streaming ecosystems is the use of publish-subscribe (“pub/sub”) messaging backbones that decouple data senders (publishers) and recipients (subscribers). Popular message-focused middleware solutions such as RabbitMQ (VMware, 2022), Apache Pulsar (Apache Software Foundation, 2022b), and Apache Kafka (Apache Software Foundation, 2022a) all provide differing capabilities as backbones. OpenMSIStream provides robust and efficient, yet easy, access to the rich data streaming systems of Apache Kafka. 

   more » « less
3. A Survey of Resource Management for Processing-In-Memory and Near-Memory Processing Architectures

   https://doi.org/10.3390/jlpea10040030  

   Khan, Kamil ; Pasricha, Sudeep ; Kim, Ryan Gary ( December 2020 , Journal of Low Power Electronics and Applications)

   null (Ed.)

   Due to the amount of data involved in emerging deep learning and big data applications, operations related to data movement have quickly become a bottleneck. Data-centric computing (DCC), as enabled by processing-in-memory (PIM) and near-memory processing (NMP) paradigms, aims to accelerate these types of applications by moving the computation closer to the data. Over the past few years, researchers have proposed various memory architectures that enable DCC systems, such as logic layers in 3D-stacked memories or charge-sharing-based bitwise operations in dynamic random-access memory (DRAM). However, application-specific memory access patterns, power and thermal concerns, memory technology limitations, and inconsistent performance gains complicate the offloading of computation in DCC systems. Therefore, designing intelligent resource management techniques for computation offloading is vital for leveraging the potential offered by this new paradigm. In this article, we survey the major trends in managing PIM and NMP-based DCC systems and provide a review of the landscape of resource management techniques employed by system designers for such systems. Additionally, we discuss the future challenges and opportunities in DCC management. 

   more » « less
4. Bridging BAD Islands: Declarative Data Sharing at Scale

   https://doi.org/10.1109/BigData50022.2020.9378342  

   Wang, Xikui ; Carey, Michael J. ; Tsotras, Vassilis J. ( December 2020 , IEEE International Conference on Big Data (Big Data))

   null (Ed.)

   In many Big Data applications today, information needs to be actively shared between systems managed by different organizations. To enable sharing Big Data at scale, developers would have to create dedicated server programs and glue together multiple Big Data systems for scalability. Developing and managing such glued data sharing services requires a significant amount of work from developers. In our prior work, we developed a Big Active Data (BAD) system for enabling Big Data subscriptions and analytics with millions of subscribers. Based on that, we introduce a new mechanism for enabling the sharing of Big Data at scale declaratively so that developers can easily create and provide data sharing services using declarative statements and can benefit from an underlying scalable infrastructure. We show our implementation on top of the BAD system, explain the data sharing data flow among multiple systems, and present a prototype system with experimental results. 

   more » « less
5. Twister2: Design of a big data toolkit

   https://doi.org/10.1002/cpe.5189  

   Kamburugamuve, Supun ; Govindarajan, Kannan ; Wickramasinghe, Pulasthi ; Abeykoon, Vibhatha ; Fox, Geoffrey ( March 2019 , Concurrency and Computation: Practice and Experience)

   Data‐driven applications are essential to handle the ever‐increasing volume, velocity, and veracity of data generated by sources such as the Web and Internet of Things (IoT) devices. Simultaneously, an event‐driven computational paradigm is emerging as the core of modern systems designed for database queries, data analytics, and on‐demand applications. Modern big data processing runtimes and asynchronous many task (AMT) systems from high performance computing (HPC) community have adopted dataflow event‐driven model. The services are increasingly moving to an event‐driven model in the form of Function as a Service (FaaS) to compose services. An event‐driven runtime designed for data processing consists of well‐understood components such as communication, scheduling, and fault tolerance. Different design choices adopted by these components determine the type of applications a system can support efficiently. We find that modern systems are limited to specific sets of applications because they have been designed with fixed choices that cannot be changed easily. In this paper, we present a loosely coupled component‐based design of a big data toolkit where each component can have different implementations to support various applications. Such a polymorphic design would allow services and data analytics to be integrated seamlessly and expand from edge to cloud to HPC environments.

    

   more » « less