The new BaBar Data Reconstruction Control System
A. Ceseracciu, M. Piemontese, F. Safai Tehrani, P. Elmer, D. Johnson, T. M. Pulliam
Abstract
The BaBar experiment is characterized by extremely high luminosity, a complex detector, and a huge data volume, with increasing requirements each year. To fulfill these requirements a new control system has been designed and developed for the offline data reconstruction system. The new control system described in this paper provides the performance and flexibility needed to manage a large number of small computing farms, and takes full benefit of OO design. The infrastructure is well isolated from the processing layer, it is generic and flexible, based on a light framework providing message passing and cooperative multitasking. The system is actively distributed, enforces the separation between different processing tiers by using different naming domains, and glues them together by dedicated brokers. It provides a powerful Finite State Machine framework to describe custom processing models in a simple regular language. This paper describes this new control system, currently in use at SLAC and Padova on ~450 CPUs organized in 12 farms.
Create a lesson
Related papers
PixelFlow: Token-Level Workload Management for Efficient Distributed DiT Serving
Zhexiang Zhang, Minchen Yu, Yifan Sun et al.
A Kubernetes-Native Request Router for Quality-Aware Inference Serving in the Computing Continuum
Ignjat Karanovic, Pantelis A. Frangoudis, Ivan Čilić et al.
Accelerating Sharded Data Parallelism at Scale with Federated Learning
Gianluca Mittone, Marco Aldinucci
Distributed Edge Inference: an Experimental Study on Multiview Detection
Gianluca Mittone, Giulio Malenza, Marco Aldinucci et al.
P-GADMM: Parallel Group-Based ADMM for Asynchronous Optimization in Heterogeneous Edge Networks
Gaiguo Wei, Qingying Zhang, Heqiang Wang et al.
VERA: Reinforcement Learning for Dynamic Memory Scaling of HPC Workloads in Kubernetes
Ade Pramono, Jie Ren, Ivy Peng