Subscribe to the gold package and get unlimited access to Shamra Academy
Register a new userJITA4DS: Disaggregated execution of Data Science Pipelines between the Edge and the Data Centre
88
0
0.0
(
0
)
Added by
Genoveva Vargas-Solar
Publication date
2021
fields
Informatics Engineering
and research's language is
English
Ask ChatGPT about the research
No Arabic abstract
This paper targets the execution of data science (DS) pipelines supported by data processing, transmission and sharing across several resources executing greedy processes. Current data science pipelines environments provide various infrastructure services with computing resources such as general-purpose processors (GPP), Graphics Processing Units (GPUs), Field Programmable Gate Arrays (FPGAs) and Tensor Processing Unit (TPU) coupled with platform and software services to design, run and maintain DS pipelines. These one-fits-all solutions impose the complete externalization of data pipeline tasks. However, some tasks can be executed in the edge, and the backend can provide just in time resources to ensure ad-hoc and elastic execution environments. This paper introduces an innovative composable Just in Time Architecture for configuring DCs for Data Science Pipelines (JITA-4DS) and associated resource management techniques. JITA-4DS is a cross-layer management system that is aware of both the application characteristics and the underlying infrastructures to break the barriers between applications, middleware/operating system, and hardware layers. Vertical integration of these layers is needed for building a customizable Virtual Data Center (VDC) to meet the dynamically changing data science pipelines requirements such as performance, availability, and energy consumption. Accordingly, the paper shows an experimental simulation devoted to run data science workloads and determine the best strategies for scheduling the allocation of resources implemented by JITA-4DS.
rate research
Read More
This paper proposes a composable Just in Time Architecture for Data Science (DS) Pipelines named JITA-4DS and associated resource management techniques for configuring disaggregated data centers (DCs). DCs under our approach are composable based on vertical integration of the application, middleware/operating system, and hardware layers customized dynamically to meet application Service Level Objectives (SLO - application-aware management). Thereby, pipelines utilize a set of flexible building blocks that can be dynamically and automatically assembled and re-assembled to meet the dynamic changes in the workloads SLOs. To assess disaggregated DCs, we study how to model and validate their performance in large-scale settings.
This paper describes how to augment techniques such as Distributed Shared Memory with recent trends on disaggregated Non Volatile Memory in the data centre so that the combination can be used in an edge environment with potentially volatile and mobile resources. This article identifies the main advantages and challenges, and offers an architectural evolution to incorporate recent research trends into production-ready disaggregated edges. We also present two prototypes showing the feasibility of this proposal.
Memory-compute disaggregation promises transparent elasticity, high utilization and balanced usage for resources in data centers by physically separating memory and compute into network-attached resource blades. However, existing designs achieve performance at the cost of resource elasticity, restricting memory sharing to a single compute blade to avoid costly memory coherence traffic over the network. In this work, we show that emerging programmable network switches can enable an efficient shared memory abstraction for disaggregated architectures by placing memory management logic in the network fabric. We find that centralizing memory management in the network permits bandwidth and latency-efficient realization of in-network cache coherence protocols, while programmable switch ASICs support other memory management logic at line-rate. We realize these insights into MIND, an in-network memory management unit for rack-scale memory disaggregation. MIND enables transparent resource elasticity while matching the performance of prior memory disaggregation proposals for real-world workloads.
Byte-addressable persistent memories (PM) has finally made their way into production. An important and pressing problem that follows is how to deploy them in existing datacenters. One viable approach is to attach PM as self-contained devices to the network as disaggregated persistent memory, or DPM. DPM requires no changes to existing servers in datacenters; without the need to include a processor, DPM devices are cheap to build; and by sharing DPM across compute servers, they offer great elasticity and efficient resource packing. This paper explores different ways to organize DPM and to build data stores with DPM. Specifically, we propose three architectures of DPM: 1) compute nodes directly access DPM (DPM-Direct); 2) compute nodes send requests to a coordinator server, which then accesses DPM to complete a request (DPM-Central); and 3) compute nodes directly access DPM for data operations and communicate with a global metadata server for the control plane (DPM-Sep). Based on these architectures, we built three atomic, crash-consistent data stores. We evaluated their performance, scalability, and CPU cost with micro-benchmarks and YCSB. Our evaluation results show that DPM-Direct has great small-size read but poor write performance; DPM-Central has the best write performance when the scale of the cluster is small but performs poorly when the scale increases; and DPM-Sep performs well overall.
The Data Activated Liu Graph Engine - DALiuGE - is an execution framework for processing large astronomical datasets at a scale required by the Square Kilometre Array Phase 1 (SKA1). It includes an interface for expressing complex data reduction pipelines consisting of both data sets and algorithmic components and an implementation run-time to execute such pipelines on distributed resources. By mapping the logical view of a pipeline to its physical realisation, DALiuGE separates the concerns of multiple stakeholders, allowing them to collectively optimise large-scale data processing solutions in a coherent manner. The execution in DALiuGE is data-activated, where each individual data item autonomously triggers the processing on itself. Such decentralisation also makes the execution framework very scalable and flexible, supporting pipeline sizes ranging from less than ten tasks running on a laptop to tens of millions of concurrent tasks on the second fastest supercomputer in the world. DALiuGE has been used in production for reducing interferometry data sets from the Karl E. Jansky Very Large Array and the Mingantu Ultrawide Spectral Radioheliograph; and is being developed as the execution framework prototype for the Science Data Processor (SDP) consortium of the Square Kilometre Array (SKA) telescope. This paper presents a technical overview of DALiuGE and discusses case studies from the CHILES and MUSER projects that use DALiuGE to execute production pipelines. In a companion paper, we provide in-depth analysis of DALiuGEs scalability to very large numbers of tasks on two supercomputing facilities.
Log in to be able to interact and post comments
comments
Fetching comments
Sign in to be able to follow your search criteria
