Skip to content

Blog

Tier 2 Storage

The University of Arizona’s Tier 2 Storage, designed to utilize Amazon Web Service storage solutions, offers scalable storage for research data that does not require immediate access. Unlike Tier 1 storage, which is integrated into high-performance computing (HPC), Tier 2 is intended for archiving inactive data or making copies. It employs AWS’s intelligent tiering, which progressively moves data to more cost-effective, long-term storage options like Glacier and Deep Glacier after periods of inactivity.

Tier 2 AWS buckets use intelligent tiering to determine the archival status of files. When data are first uploaded to a group's bucket, they are in the standard access class. After three months of inactivity, data are automatically migrated to Glacier storage. This is less performant, and data are no longer instantly downloadable. After three months of inactivity in the Glacier access tier, data are automatically migrated to Deep Glacier, which is even less performant. Users will need to request a restore before downloading their files from Glacier and Deep Glacier. Restore requests can be submitted either in the user portal or using a command line tool available on our compute nodes.

Did I mention that the service is highly subsidized? You will set up an account to cover the costs of the data that is temporarily in S3, but there are no charges for data migrated to Glacier and Deep Glacier. And there are no ingress or egress charges.

For detailed usage and pricing information, visit the official Tier 2 Storage page.

A New Era in Research: The Nobel Prizes Showcase AI’s Transformative Power

What do two of the recent Nobel Prize awards have in common? In both cases of the Chemistry award and the Physics award, the committees recognized the transformative power of artificial intelligence, and the high-performance computing (HPC) that underpins it.

Geoffrey Hinton of Google DeepMind and Princeton professor John J Hopfield received the physics honor for their groundbreaking work in artificial neural networks. “The laureates’ work has already been of the greatest benefit. In physics we use artificial neural networks in a vast range of areas, such as developing new materials with specific properties,” says Ellen Moons, Chair of the Nobel Committee for Physics.

Demis Hassabis and John Jumper from Google's AI division DeepMind and David Baker from the University of Washington were awarded the 2024 Nobel Prize in Chemistry. Hassabis and Jumper received the award for AlphaFold2, an AI system that accurately predicts the 3D structures of proteins from their amino acid sequences in minutes. AlphaFold has predicted over 200 million protein structures, and has, so far, over 2 million users. This means it has already potentially saved millions of dollars and hundreds of millions of years in research time.

Researchers at the University of Arizona take advantage of AlphaFold to advance their studies of protein folding using Puma, our newer supercomputer. We host the dataset to save researchers a lot of time, storage capacity, and provide compute performance. Our copy has over 200,000 files in 2.8TB. The latest is a lot bigger. We also host the containers with the code required.