Skip to main content


For individuals, there is no charge for user accounts on the traditional compute cluster. These are paid for by your research group.

The Jupyter cluster is free for all users thanks to a Vanderbilt TIPs grant.

ACCRE is supported by use and support fees from its research groups. We allow new groups to try out ACCRE as a guest group to ensure that it meets their needs, and we will only ask for fairshare after several months of sustained usage.

All services are on a fiscal year commitment and billed monthly. Reduced rates for research groups are subsidized by VU and VUMC.

Last updated May 13, 2021.

Cluster Support and Use Fees

ServicePrice to ResearcherUnsubsidized Price
Cluster Support Fee (ACCRE will purchase a node for your group)$480/node/year$863/node/year
Cluster Use Fee (your group rents a node from ACCRE)$84/core/year$88/core/year
  • Per node service support fee payments require upfront purchase of cluster servers (i.e. compute nodes) approved and ordered by ACCRE but paid for directly by research group plus monthly support fee payments.
  • Per node fairshare is equal to the number of CPU cores in the purchased hardware, and is applicable for the duration of the purchased hardware warranty. Current compute nodes (as of June 2018) contain sixteen CPU cores.
  • Fairshare is ACCRE’s computing “currency” for groups on the cluster. One CPU core of fairshare equates to 24 hours of CPU time over the course of a day, or 365 * 24 = 8760 CPU hours per year. More information about fairshare can be found here.
  • A group’s bursting capacity is generally well above (by a factor of 5 – 20) its group fairshare. In other words, a group paying support fees on sixteen CPU cores may be allowed to burst up to several hundred CPU cores on the cluster at any given time, with the idea being that the group’s average usage over the course of several weeks or months should be roughly equal to that of the group’s fairshare.
  • Use fees can be paid for computing on a per-CPU-core basis. This does not require the purchase of hardware upfront, but does require a minimum of a seven month commitment.

GPFS Filesystem

ServicePrice to ResearcherUnsubsidized Price
/data (backed up nightly, recommended)$204/TB/year$217/TB/year
/scratch (not backed up)$150/TB/year$159/TB/year

GPFS is filesystem software used for achieving high-performance I/O on the cluster. Technical details can be found here.

Custom Gateways

ServicePrice to ResearcherUnsubsidized Price
  • Custom gateways require upfront purchase of server approved and ordered by ACCRE but paid for directly by research group plus monthly gateway support fee payments. Click here for more information on custom gateways.
  • Depending on the needs or requirements (i.e. the support burden) of a group’s custom gateway, it may be classified as either basic, intermediate, or complex. Please refer to the customer gateway service agreement for more specifics.

Tape Backups

ServicePrice to ResearcherUnsubsidized Price
Internal Server$180/TB/year$442/TB/year
External Server$240/TB/year$293/TB/year

Tape backups are available for both internal (on ACCRE network, typically for customer gateway servers) and external (i.e. not on ACCRE network; e.g. a server or workstation in a group’s lab space) servers.

SciLo (ACCREx)

ServicePrice to ResearcherSubsidized Price
SciLo Tape Storage$49.98/TB/yearN/A*

Groups requiring a user-managed archive solution may wish to consider ACCRE’s tape-based SciLo Service.

User-managed means that users, not ACCRE, are responsible for performing the backup and restores of their data by creating their own backup schedules or manually initiating operations. Tape-based means users should not expect this service to behave like a disk-based storage system. Reads from SciLo can experience some delay while the tapes needed to perform a restore are being loaded.

The SciLo Archive is accessible while on campus and over the public Internet. The user can archive files from as many computers (servers and laptops) as they want.

*As an ACCREx program, SciLo is not eligible to be subsidized.