Tape Storage

Tape storage gives BYU researchers an inexpensive, easy-to-use place to keep large datasets for the long term. It is designed for data that matters but does not need to stay on fast storage all the time. Data transfers use Globus, so moving data into or out of the archive is straightforward and familiar.

Tape storage is a good fit for large research datasets, completed projects, raw instrument output, and other data you need to retain but do not access every day. It is not intended for active scratch space or for data that moves back and forth constantly during normal work.

This service is only subsidized for the research needs of BYU researchers.

Why use tape storage

• Subsidized one-time price of $1000 per 30 TiB ($909.50 per 30 TB) for 7 years
  • 1 TiB ≈ 1.0995 TB. OS tools usually report TiB (even if labeled "TB"); vendors market TB because it appears larger.
• Two copies of all archived data, with the replicated copy stored in a different building on campus
• Simple transfers through Globus

This service is only subsidized for the research needs of BYU researchers.

Best uses

• Large datasets that are rarely accessed
• Long-term retention of completed or published project data
• Raw data that must be preserved
• Reference data you only need occasionally
• Large datasets you occasionally want to move to or from collaborators using Globus

This service is only subsidized for the research needs of BYU researchers.

How it works

This service uses Versity's ScoutAM software behind the scenes, but from the user perspective the main experience is simply transferring data with Globus.

From your point of view, the tape storage appears as another storage location you can use with Globus. You transfer data into the archive when you want to keep it for the long term, and you transfer it back out when you need it again.

This solution includes a large disk cache. If your data is still in cache, it is served directly from disk and does not need to be recalled from tape. If it is no longer in cache, the archive recalls it from tape automatically.

Because the archive is tape-backed, retrieval is not as immediate as active disk storage. Recalls from tape typically take a minute or two to get started, although they can take longer if all tape drives are busy at the time. Most of this latency is hidden from you since Globus is designed to be a fire-and-forget data transfer solution.

All data is replicated to a second tape library which is located in a different building on campus. The second copy is created asynchronously, so replication may complete hours or even days after the first copy is written.

Storage is only sold in multiples of 30 TiB.

This service is only subsidized for the research needs of BYU researchers.

Proper usage

Tape storage works best when used as an archive, not as active working storage.

Keep these guidelines in mind:

• This heavily subsidized tape solution is subsidized for BYU research needs, not for anything else.
• Do not use it for data that moves back and forth all the time.
• Do not treat it like live storage for ongoing analysis.
• Avoid millions of very small files. (e.g. create a tar file with the files)
• Prefer fewer, larger files whenever possible.
• Bundle small files before transferring them when it makes sense (e.g. create a tar file)
• Organize data before archiving so it is easy to find and retrieve later.

A simple rule of thumb: if you expect to work with the data regularly, keep it on active storage. If you need to keep it safely and retrieve it only occasionally, tape storage is the right fit.

This service is only subsidized for the research needs of BYU researchers.

Data transfers

Globus handles transfers to and from the tape storage through a web-based interface. That makes it easy to move data between BYU systems, personal systems, collaborator institutions, and the archive without needing to manage tapes directly.

For archive workflows, Globus is especially useful because transfers are reliable and do not require you to sit and watch them. You submit the transfer, and Globus handles the movement in the background.

NOTE: POSIX permissions and ownership are not preserved by Globus. When you transfer the files back to disk, the files are created with the permissions and ownership of the person who is performing the transfer.

Availability

July 27 update: The system is now available by invitation based on previous surveys. We will reach out to specific groups for the next few weeks then open it up generally after that.

Getting started

This service is only subsidized for the research needs of BYU researchers.

For more information or to get started, open a support ticket or email rcsupport@byu.edu. Let us know:

• How much storage you need (only sold in multiples of 30 TiB)
• A brief description of the data and how you will use it
• Start date
• (If known) Which file sharing group to associate with the data

Terms

Before the end of the service term, all data must be removed, transferred to another storage location, or renewed for an additional term if renewal is available. Because this service is heavily subsidized, all payments are non-refundable, including in cases of early termination.

Access associated with an individual account ends when that person leaves the university. Before departure, ownership of the data must be transferred to another authorized user, or the data must be removed.

All applicable Office of Research Computing and BYU policies remain in effect.

Side note

It might seem confusing to have a disk solution named "archive" then introduce a tape tier at a later date, which is itself more of a true "archive" than the disk solution. Oops. 😬