Grad Net Static Datashare

INTRO

There might times where, for a class or for your light research, you might need to download or import a large set of data (either very big files or a large number of files), where that data is static, i.e, will not be modified. The problem is that when you download very large static files or a data set with lots of static files to your Grad Net home directory, you are using a significant amount of disk resources shared by the much of the CS Dept, especially, the backup resources our systems use. (Backing up very large files or lots of smaller files puts a load on our storage/backup resources.)

If you're storing static files, then, they should not have to be backed up or mirrored (as we do with our home directories), as you can re-obtain those files when necessary from your original sources. Therefore, we have created an alternative... a Grad Net static datashare.

NOTE: Both the Grad and UGrad nets use a different, special data storage location for static data specifically provided from CS professors to their classes for course assignments (and that's usually not big data.) And that should not be confused with the Grad Net static datashare described on this page. So, if you're teaching a course and need to provide static data to the students, that is a different situation, and you should please email support@cs.jhu.edu to arrange access.

About the Grad Net Static Datashare

The Grad Net Linux clients have a secondary location for data storage to supplement the primary storage afforded by everyone's home directories. That secondary storage is available under the /static directory.

That /static directory is the same across all of the Grad Net Linux clients. So, if you place files in /static on one Grad Net Linux client, it will be available in the /static folder of each of the Grad Net Linux clients.

The /static directory is intended for data that:

• Is larger than a few gigabytes in size
  • Consider anything 10gb and over in file size to be large
• Could instead be a large number of files.
  • Consider a data set with more than 1,000 files to be large (regardless of filesize)
• Could be a complete dataset/folder with a moderate number of data files, but whose combined file sizes make it a large folder.
  • Consider a dataset or folder consisting of a reasonable number of static files in that folder, but, where the total size of the data in the folder is 10gb or more, to be large.
• Does not need to be backed up
  • It's static data that you acquired from somewhere, so you can obtain it again if necessary.
• Needs to be available to all CS Grad Net Linux clients
  • /static is shared amongst the Grad Net Linux clients
  • NOTE: Static datashare storage is not available for the Ugrad Net
• Should not be deleted by automatic system cleanup processes

For example, if there's a 30 gigabyte data set you want to download and look at, you could use /static to hold the data while you're working with it. You would put any programs you write to analyze the data in your home directory. If something happened to the department's servers, we would restore your programs from our backups (since we back up the home directories), and you could re-download the data from wherever you got it originally (since we do not backup the static datastore.)

Therefore, putting your static data into /static helps the department because it limits the expense of backing up excess data unnecessarily.

How to Use the Static Datashare

First, log into one of the Grad Net Linux clients.

Then, simply create directory in /static to hold your data. We suggest (but don't require) that you use your account name. So someone with an account name of "jdoe15" could, for example, run:

mkdir /static/jdoe15
cd /static/jdoe15
wget https://example.com/some/dataset.zip

Our www.cs.jhu.edu website can also see your /static data share!

In addition to being available on the Grad Net Linux clients, /static is also available on the department webserver, so you can have PHP/CGI programs that make use of data in /static. (And if you want to serve files in /static directly, you can just put symbolic links to them from somewhere under your ~/public_html directory.)

Limits on Static Datashare Use

The usual restrictions on use of CS Department resources apply; any data stored in /static should be related to the department's educational or (light) research activities.

This service is only available to people with CS Grad Net accounts. There is no analogous service for people with only a CS Ugrad Net account.

How much data can I store in /static?

Every CS Grad Net account has a quota to limit overuse of the /static system. The default user quota for /static is 50 gigabytes. Currently, we do not have a way for you to examine your /static disk quota, so, you will need to keep track of the amount of data you have stored there. Also, in /static, the number of files is not an issue, as long as the total size of all your files in /static is below your quota.

If you need to store more data in /static than the default quota allows, please email us at support@cs.jhu.edu with a description of what you're storing and how much total space you expect to need. (Quota increases are not guaranteed.)

NOTE: CS Grad Net home directories do not have disk quotas. Quotas on the CS Grad Net only apply to the /static directory.

Backups

Although the data in /static is not backed up via our usual mechanisms, there is a limited self-service backup system similar to our home directories' Self-Service File Recovery. See that page for instructions. Note that if self-service file restoration does not work, you will have to download or recreate the dataset yourself.

Alternatives to the Static Datashare

There are several other places where you can put files, depending on your needs:

• Files in your home directory are available on all of our client systems that share the account. Such files are also backed up as described on our Backups page.
• Each Linux client has a /scratch directory. Files put there are only accessible on the Linux client where they're stored. Not only are the files there not backed up, they will be automatically deleted during our annual client OS upgrade process (and may be deleted at other times as circumstances warrant).
• Each Linux client also has a /tmp directory. Like /scratch, files put in /tmp are only accessible on the Linux client where they're stored. Files in /tmp are automatically cleared out much more frequently than files in /scratch.