UVA Research Computing provides multi-tiered storage solutions for your data. From the individual cloud storage solution for your personal computing to the high-performance parallel file system for serious computational runs, various storage solutions are available for your need. In order to help you find the right storage, let’s start with one simple question:
Is your data research data?
If yes, please check out the research computing storage solutions. The primary purpose of these storages is to provide adequate space for your large-scale research data. This includes file systems for Rivanna HPC system and long-term storage for large-scale data as well as secure storage for sensitive data.
If not, ITS provides an array of enterprise storage solutions that comes with multiple layers of data protection (snapshot, replica, and tape archiving). The minimum request for storage space starts from 10GB, it may suite you well if you are dealing with small size of very important data.
UVA Box is a cloud-based storage and collaboration service that gives eligible members of the University community the ability to access, store, and share up to 1 TB of non-sensitive/moderately sensitive University files securely—anywhere, anytime, on any device.
CrashPlan is a cloud-based desktop backup service. It securely backs up your endpoint devices to the cloud. CrashPlan provides:
- Cloud storage for backup of up to 4 endpoint devices per user
- Protection against crypto-ransomware and other malicious software that destroys/encrypts content on end-user’s devices
- Protection of University data on endpoint devices from loss due to hard drive failure, computer failure, etc.
CrashPlan is currently offered at no cost to the University community until June 30, 2018. After that, the cost model/fee structure will be determined for continued use of the service. During this this initial phase, the system has a per-user quota of 250GB. If you need more space and have a valid use case, please contact ITS via the link below.
High Performance Computing
/home File System (FREE)
Each user on Rivanna HPC cluster is provided with 50GB of
/home directory. It is a standard place where you can store important files or data such as your research code, configuration files, and valuable output data. Users can compile, debug their codes in this space before getting ready for production runs via scheduler (SLURM on Rivanna) on compute nodes. The
/home comes with 3 weeks of snapshot backup, so if you delete your data by mistake, you can ask file recovery. The 50GB is the hard quota of the space, and users are not allowed to exceed this limit. You job will fail if the hard limit is reached.
/scratch File System (FREE)
/scratch file system is a large-scale, high-performance parallel file system (Lustre) where multi-threaded, high-performance read & write is possible. The scratch space is freely provided, and one of the most versatile storage locations available to you -- for certain use cases and with the right tuning,
/scratch performs extremely well: upwards of 10x faster than NFS-mounted storage. We recommend to use this file system whenever running a series of jobs on RIvanna. There are quotas imposed (10TB per user), but they are not strictly enforced: being over quota usually just means not being able to submit jobs until you're not over quota any more, and in many cases you can have your quota adjusted simply by asking.
/project File System ($90/TB/YR)
/project storage option provides storage for collaboration and data sharing within the research group. Like
/scratch file systems,
/project is mounted on Rivanna, and users can freely move data around between those file systems. Total usable space of
/project is close to 2 PB, and the file system comes with a data protection of 3-week snapshot. Non-Rivanna users can still request the storage space, and the data transfer to/from the storage is possible via a Globus data transfer node (More details here). UVA faculty can purchase
/project space by submitting this form.
Research Value Storage ($45/TB/YR)
Research Value Storage is a low-cost, moderate-performance version of the Enterprise Storage offered by ITS. Users can request a space on this system by submitting this form, and can use this storage without having to get an account Rivanna HPC system. It can be mounted from the central network by clients on local laptops and workstations running Linux, Windows, or Mac OSX. Although not as fast as other UVA storage solutions, Value Storage offers a familiar format which is easy to understand and use. It is differentiated from Enterprise Storage by the lack of data protection features and services that are available on the Enterprise tiers.
Academic Value Storage ($90/TB/YR)
ITS offers storage for non-research users for hosting their non-sensitive data. Users can request a space on this system by submitting a form. It can be mounted from the central network by clients on local laptops and workstations running Linux, Windows, or Mac OSX. Although not as fast as other UVA storage solutions, Value Storage offers a familiar format which is easy to understand and use. It is differentiated from Enterprise Storage by the lack of data protection features and services that are available on the Enterprise tiers. It is good for users who have their own backup solutions in place, and need a low cost terabyte level storage solution
Storage Solution for Sensitive Data
Ivy Secure Storage ($TBD/TB/YR)
UVA’s secure data analytics system, Ivy, has multi-tiered storage systems, and a PI can specify the storage space s/he would like to have when requesting access to Ivy. A Central Storage pool of more than 1 PB of space is available for highly sensitive data storage regardless which platform users select for their project (among three available platforms: Virtual Machines, Domino DataLab, and Hadoop/Spark). For more information, please check out our Ivy User Guide. This storage doesn’t allow users to store executable files for security reasons, and provides read & write only access. Executable files can be stored on the VM storage.
Virtual machines do not come with any significant disk storage of their own, and small amounts of data (less than 100GB) can be stored directly in the block storage space that comes attached with each VM. If larger storage space is required, for relatively higher I/O performance than the central storage space, users can request a space on the network attached storage (NAS) mounted on the VM.