Abstract
The CERN OpenStack cloud provides over 300,000 CPU cores to run data processing analyses for the Large Hadron Collider (LHC) experiments. To deliver these services, with high performance and reliable service levels, while at the same time ensuring a continuous high resource utilization has been one of the major challenges for the CERN cloud engineering team. Several optimizations like NUMA-aware scheduling and huge pages, have been deployed to improve scientific workloads performance, but the CERN Cloud team continues to explore new possibilities like preemptible instances and containers on bare-metal. In this paper we will dive into the concept and implementation challenges of preemptible instances and containers on bare-metal for scientific workloads. We will also explore how they can improve scientific workloads throughput and infrastructure resource utilization. We will present the ongoing collaboration with the Square Kilometer Array (SKA) community to develop the necessary upstream enhancement to further improve OpenStack Nova to support large-scale scientific workloads.
Highlights
During the last 5 years, CERN has been running a large cloud infrastructure that provides an IaaS service to the CERN community
At same time the Square Kilometre Array (SKA) will face similar challenges to process the huge amounts of data from its high sensitive radio telescopes
In this paper we will focus on the CERN cloud infrastructure use-case
Summary
During the last 5 years, CERN has been running a large cloud infrastructure that provides an IaaS service to the CERN community. The cloud infrastructure brought several advantages when compared to the previous model, the manual allocation of physical resources. It improved the user responsiveness with a self-service kiosk for virtual resources, enabled cloud interfaces like OpenStack APIs and EC2 API, improved efficiency over the entire lifetime of the resources [1]. Preemptible instances are deleted as soon a tenant tries to create virtual machines within its quota. The paper is organized as follows: Section 2 presents a summary of the CERN cloud architecture including a brief description of the OpenStack project.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have