CloudyCluster Team's blog
Cloud HPC Insights
© 2020 - 2022 CloudyCluster GCP Blog
A Box in the Cloud for LAMMPS
Introduction One of the great features of LAMMPS is that it can run on single processors or in parallel using message-passing techniques and a spatial-decomposition of the simulation domain. As it’s named acronym implies, Large-scale Atomic/Molecular Massively Parallel Simulator is well suited for parallelization in the cloud, and this article will review ways to optimize those simulations. This can work equally well for Molecular Dynamics and Materials Sciences simulations.
Background/Problem Statement (Boundary 3 - x, y, z) If you were to take one of the example input files from the LAMMPS installation directory, you can run the simulations in a few seconds or minutes on a local or cloud machine.
Cloud HPC Hackathon @ SC19
Denver, CO – November 2019 marked another successful Super Computing conference, but for SGCI and the Omnibond CloudyCluster Team, it was even more special as 5 teams competed at the Cloud HPC Hackathon at SC19. Teams consisted of students from universities at all levels, ranging from Freshman to Ph.D. candidates working together under pressure of the clock and pop-up challenges over a 44 hour period.
Schools represented at this year’s event included locals from University of Denver and University of Colorado.
How API-driven Public Clouds and HPC are Bringing a Brighter Future
(reprint)…
API-Driven Public Cloud The increasing popularity of cloud computing is evident in a myriad of news articles, blog posts and videos. Although cloud computing offers many benefits, one of the greatest benefits is largely hidden within the cloud.
When public clouds appeared, initial conversation focused on comparing the public cloud to data center virtualization. A critical aspect gradually emerged, the secret sauce: the public cloud is more than just virtualization.
Introducing CloudyCluster on Google Cloud Platform
Many customers have been asking how to seamlessly get HPC jobs to work between clouds and on premise. We are now making this a little more possible by extending CloudyCluster to the Google Cloud Platform. It is avaialble now by request and will be in the GCP Marketplace in a few weeks. Below is the press release: Enjoy…
DALLAS (PRWEB) NOVEMBER 12, 2018
Today, Omnibond announces the availability of CloudyCluster on Google Cloud Platform (GCP).
Leveraging GCS Storage for Efficient Access to Large Data Sets
Data sets for investigation and research can be critical and need to be kept around for at least the life of a research project, but possibly much longer based on data retention policies. Google Cloud Storages (GCS) is object storage that is resilient and easy to use and scale. GCS has such features as:
*>*9.99% availability for standard, and *>*99.95% for other storage classes Multiple storage classes (Standard, Nearline, Coldline, and Archive) which can be managed automatically based on access frequency or time.
Running Science Gateways on CloudyCluster
CloudyCluster creates an HPC environment in the Public Cloud that is similar to on premise HPC systems. To provide elasticity in the cloud, we have developed a meta-scheduler. The CCQ meta-scheduler enables jobs to control elastic launching and deleting of instances as required. These can be diverse instance types including preemptible and standard as required by the individual jobs.
To support the SGCI Cloudify Gateways solicitation https://sciencegateways.org/cloudify2019 we will be adding new simplified HPC Job Submission APIs for CCQ to the next release of CloudyCluster targeted for the SC19 timeframe.
The OrangeFS Project, an Open Source Product with a Research Outreach
(reprint)…
More and more businesses rely on data-driven decisions, simulations, models and customer interaction. More of these data-driven requirements need faster and faster responses, even real-time, and this demand is stressing traditional storage architectures. Historically, storage systems and protocols have single points of ingress and egress for data. These single pinch-points are increasingly becoming bottlenecks, limiting the ability to process data in a timely fashion.
The High Performance Computing community has faced these problems since the inception of the Beowulf cluster, the original concept of tying several commodity computers together to distribute computational workloads.
WRF Parallel Modeling
Parallel Weather Modeling in the Cloud
Weather and climate modeling is becoming more critical as more and more industries rely on accurate forecasting from typical weather to severe storms. This modeling can provide insights and cause/effect relationships we have yet to discover. It can also extend our current understanding of the changing climate.
In this experiment, the computation tool is the Weather Research and Forecasting (WRF) model. WRF is a next-generation mesoscale numerical weather prediction system designed for both atmospheric research and operational forecasting applications.