Research Data Storage

Data Storage on the HBSGrid

HBS provides storage for active research projects. Research storage is typically used in conjunction with the computer cluster (HBSGrid) and accessed via NoMachine remote desktop as documented in the Userguide.

There are three sets of storage locations available for research work, depending on the particular usage patterns, size consideration, and number of people involved in the work. These are the home folders, scratch folders, and project spaces.

Folder Type	Size	Expandable?	Shareable?	Backed up?	Other Considerations
Home	150GB (100GB for guests)*	No	No	Yes	This is a personal folder with size limitations that cannot be shared with others.
Scratch	Varies	Yes	Yes	No	Files older than 60 days are deleted and files are not backed up. This is SSD storage that is faster than other storage options.
Project	Default is 50GB	Yes	Yes	Yes	These folders are meant for collaboration and/or projects that may increase in size over time.

*Default sizes for home folder have grown over time. If you received your HBSGrid account prior to 2024, you may have a smaller home directory

Home folders

By default, when your account is created (whether this account is used for just storage or storage + computing), a home folder is created. Home folders are 150GB (or 100GB for guests) and cannot be expanded (note: default sizes for home folder have grown over time. If you received your HBSGrid account prior to 2024, you may have a smaller home directory).

If you are logging in to the HBSGrid to do work via the NoMachine GUI or terminal, you are automatically placed in and are using this home folder. This is located at:

/export/home/<group>/<username>

For example faculty member John Harvard's home folder would be at:

/export/home/faculty/jharvard

When space in your home folder fills up, you will not be able to do any more work, which may lead to programs acting strangely or crashing altogether, disk error notices, or input/output errors. Keep an eye on your space usage and periodically remove any old files that are no longer needed.

Home folders are backed up every night. If you should need to recover any files due to accidental deletion or corruption, please contact RCS.

Disk Quota Exceeded Error

WARNING
Reaching your home folder limit can prevent NoMachine sessions from starting, and this is one of the most common reasons for difficulties connecting to the HBS Grid desktop via NoMachine.

You can fix this problem yourself in a few ways:

By Terminal: Open a terminal (in the Windows search toolbar, type "Cmd" or "Windows PowerShell"; in the Mac search toolbar, type "Terminal") and run ssh <username>@hbsgrid.hbs.edu (replace <username> with your actual HBS Grid username). Once connected you can use terminal commands like ls to list files in the directory, rm (remove) plus the name of the file to remove files you don't need, or mv (move) plus the name of your file and a path to a new location to move files. Removing and moving files can help get your home directory back under your storage quota. You can also run gio trash --empty to empty the trash, which may give you enough breathing room to permit NoMachine login.
By GUI: Use a GUI SFTP client like FileZilla or CyberDuck to log into your storage and clean up your home directory by moving/deleting files and emptying your trash folder.

Scratch storage

Whether you are doing batch or interactive work, at times you may need a temporary location to stash files that you will not keep, or your software may require a 'working', or temporary, directory. In most cases, you should not use your home folder, as there are quota limits, and the temporary needs may exceed the amount available.

The storage location at /export/scratch is designed specifically for this purpose. See the next section for setting up and using a temporary folder on this volume.

Nota bene:

This volume is a shared, community area. Be mindful of your usage (how much you use and for how long).
This filesystem is not backed up! Ensure you copy anything important back to your home or project folder.
Files older than 60 days will be deleted nightly via disk cleanup scripts. Do not interfere with these scripts.

Please see our RCS Policies page for more information about our scratch usage policies.

Using `/export/scratch` effectively

As this volume is a shared area visible by everyone, it is important that you follow best practices for its use:

Do not store any files in the top-level.
Use the Create Scratch Folder script to create your temporary folder. This will open a terminal / text window that will guide you through the process of naming your folder and setting the correct permissions. For folder naming, we recommend prefixing the folder with either your account name or the project name. To start:
- In the NoMachine GUI, select from the menubar Applications > Other > Create Scratch Directory.
- In terminal, execute the command /usr/local/app/scripts/create_scratch_folder.sh
Clean up files when you no longer need them, so that others can use the additional space immediately. Please note that this includes undeleted trash files on scratch. Although your trash may appear empty, 'trashed' files may still exist in a hidden folder and take up space.

If you should need to keep files on the scratch volume for longer than 60 days, please contact RCS.

Undeleted Trash files on Scratch

Although your Trash may appear empty, 'trashed' files may still exist in a hidden folder on scratch. To check whether this is the case, run ls -al /export/scratch and see whether a .Trash folder with your username is listed.

We recommend that you review and 'delete permanently' the files in your NoMachine/Gnome Trash. If you are still uncertain or your Trash folder is empty, you can delete your "trash" folder on /export/scratch with one of the following three terminal commands:

rm -ri /export/scratch/.Trash-$(id -u $USER) # to be prompted for every file
rm -rI /export/scratch/.Trash-$(id -u $USER) # to be prompted, but less intrusive
rm -rf /export/scratch/.Trash-$(id -u $USER) # NO PROMPT

Project spaces

Project spaces (folders) are the primary, recommended location for storing and doing collaborative work on research storage, including HBS and guest user accounts. This is in contrast to home folders, which are accessible only by the account holder.

The default size is 50GB, with increases granted upon request and with space available. Keep an eye on the space usage, as project spaces that reach capacity will throw errors in programs and when transferring files, and data loss may result.

Project spaces are backed up every evening. If you should need to recover any files due to accidental deletion or corruption, please contact RCS.

Requesting a project space

Project spaces can be set up with faculty sponsorship and/or approval by filling out the New Project Space Request Form.

If you are working with level 3 data or higher, we will ask you to submit documentation from the IRB or a Data Usage Agreement (DUA; a sample can be found here) so that we can ensure that the project space's security levels are appropriately set up. If you are unsure what level your data falls under, please review this page on data security.

If a project space already exists and you would like to make changes to the space users or size, fill out the Project Space Change Request Form.

Archiving a project space

Project space usage is reviewed and confirmed on a yearly basis. Inactive project spaces will be backed up and archived only after contacting the primary faculty sponsor. Archives are retained for the period of time specified by HU or HBS data retention policies, whichever is longest. If you would like to unarchive a project space, contact RCS.

For information about how to archive your data to an external drive, see our technical note on archiving your research files.

SQL Databases

RCS hosts a database server running MariaDB (an open source fork of MySQL) to help meet growing data storage needs. To obtain a database account, please fill out the New MariaDB Account Request Form. Please note that RCS provides limited support for databases.

Connecting to your Database

Please contact us at research@hbs.edu for connection parameters including USER, HOSTNAME, and CA Certificates. Note that some clients (including DBeaver) require you to set the SSL CA Certificate. Connections from outside the HBS Grid always require an RCS-provided SSL CA Certificate.

Configuration

Most MySQL clients will read connection information from a configuration file found in ~/.my.cnf. This file is used to store connection details to the MariaDB server, such as account details and connection parameters. If you don't have this file (on the Grid) you may request a template or create your own with the following details:

Important

A configuration file with all connection details enables database access by anyone with access to the file. Ensure that this file is not shared or accessible by anyone but you.

[client]
host=HOSTNAME
port=3306
ssl-ca=PATH_TO_SSL_CERT
database=jharvard
user=jharvard
password=PASSWORD

Please note that you will substitute jharvard and PASSWORD with your MariaDB username and password. For the other parameters such as HOSTNAME, please contact us at research@hbs.edu.

Once your .my.cnf is ready, move it your home directory and adjust file privileges to ensure no one else can read the file. On the Grid or on a Linux or Mac local machine, you may prevent others from reading the file by opening a Terminal and running this command:

chmod 700 ~/.my.cnf

On a Windows machine, right-click on the file, select Properties, and then:

Under the General tab, toggle ON the Hidden attribute
Under the Security tab, ensure only the following users or groups have access: System, Administrators, and you.

Tools to Connect to your Database

You can connect to your database using any compatible client. If you already have one you like, feel free to use that. Otherwise we recommend one of the clients listed below.

WARNING

If you are connecting to MariaDB on the cluster, please be advised that there is a known bug in the current environment. If you'd like to connect to the database using software on the cluster (e.g., R, Python), please select the rcs/rcs_2022.11 software environment version in the interactive launcher or in Terminal. If you'd like to use DBeaver, launch a Terminal session, load the the previous environment (ml load rcs/rcs_2022.11), and then once loaded launch DBeaver by typing dbeaver in the Terminal.

PythonRDesktop and DBeavermycli and TerminalODBC

Use connector-python to connect following the official documentation. It is recommended to use connection settings from ~/.my.cnf as described above.

Use RMariaDB or dbplyr, both use connection settings from ~/.my.cnf as described above. Sample code for connecting R to the Grid's MariaDB database can be found here.

Use DBeaver to connect following the official documentation. Make sure to set the CA Certificate path in the SSL connection settings tab. The DBeaver client may prompt you to download a driver to connect to your database. This is usually safe, and drivers will be stored in your home directory, under ~/.local/share/DBeaverData/drivers

The mycli client uses connection settings from ~/.my.cnf as described above. You can download mycli onto your local machine, or use the instance already installed on the Grid. To connect from your local machine type: mysql -u username -p -h mariadbhost --ssl-ca=/path/to/file/ca-cert.pem or from the Grid: mysql -u username -p -h mariadbhost.

Alternatively, you can access MariaDB from Terminal after first logging into the Grid (see instructions on the Start Here page and then typing: mysql -h hostname -u username -p.

For all of the commands referenced above, replace the username with your username, the mariadbhost with the HBS host name in your .my.cnf file, and (if applicable) the ssl-ca path with the appropriate path. You will be prompted to provide your password.

Setting up an ODBC connection differs by whether you are connecting from the Grid (Linux), Mac, or Windows machine. Please see MariaDB's ODBC set-up documentation for more information and note that if connecting from the Grid, the ODBC administration tool and appropriate ODBC drivers are already installed. You will still need to ensure that you have a .odbc.ini configuration file saved in your home directory. An example .odbc.ini template is below

[ODBC Data Sources]
rcs_mariadb = RCS_MariaDB_ODBC_connection

[rcs_mariadb]
driver = mariaDB
server = HOSTNAME
port = 3306
database = jharvard
user = jharvard
password = PASSWORD
sslmode = required
sslca = PATH_TO_SSL_CERT

Once your .odbc.ini is ready, save it to your home directory and adjust file privileges to ensure no one else can read the file:

chmod 700 ~/.odbc.ini

Importing Data

The following is a basic overview of the import process. Complete documentation for the LOAD DATA command can be found at https://mariadb.com/kb/en/library/load-data-infile/.

Move your data to an import folder
Create the database table that will hold the imported data
Import the data
Validate the import
Remove your data from the import folder

To create the database table (step 2), you will need to know the name of all columns, each column’s data type (integer, numeric with decimals, string of characters, etc), and each column’s maximum width. For example, if one of the columns in your data is US phone numbers of the format 6174953292, then you may opt to use int(10) as the column data type (i.e., integer with up to 10 digits). However, if you suspect some entries have dashes such as 617-495-3292, then you will need to use char(12) which stores the data as a string of characters, up to 12 characters in length.

Import Example

We will use the following to illustrate importing:
MariaDB username = jharvard
MariaDB database = jharvard_database
MariaDB database table = table_import
Import filename = SampleData.txt

Data:

COLUMN_1	COLUMN_2	COLUMN_3
25	Harvard	Way
86	Brattle	St

1. Move your data to the appropriate import folder

Prepare an import folder and ensure it has the appropriate permissions:
mkdir /export/mdb_external/import/jharvard
chmod 700 /export/mdb_external/import/jharvard

Move your data to this import folder:
mv SampleData.txt /export/mdb_external/import/jharvard

2. Create the database table that will hold the imported data

Log into MariaDB:
mysql -h HOSTNAME -u jharvard -p

Create a database table that will hold imported data, specifying the maximum size of each column. For example, below we specify that all columns are char with a maximum length of 20. You can modify your table at a later time, e.g., switching from char(20) to char(30):
use jharvard_database; create table table_import (Column_1 char(20), Column_2 char(20), Column_3 char(20));

3. Import the data:

load data local infile ‘/export/mdb_external/import/jharvard/SampleData.txt’ into table table_import fields terminated by ‘|’ lines terminated by ‘\n’ ignore 1 lines;

The command above has 4 sections:
(1) load data local infile ‘/export/mdb_external/import/jharvard/SampleData.txt’
specify file to import
(2) into table table_import
specify table that will hold the imported data
(3) fields terminated by ‘|’ lines terminated by ‘\n’
specify delimiters (the character which splits data or text into separate fields). More information about delimiters can be found here: https://mariadb.com/kb/en/delimiters/.
(4) ignore 1 lines
include this only if your file includes column header information

Official documentation for this command may be found at https://mariadb.com/kb/en/mariadb/load-data-infile/.

4. Validate the import

Do a preliminary check on the first 10 rows of our data:
select * from table_import limit 10;

Log out of MariaDB:
exit;

5. Remove your data from the import folder

rm -rf /export/mdb_external/import/jharvard

Exporting Data

The following is a basic overview of the export process followed by an example. Complete documentation for the SELECT … INTO FILE command to be run within MariaDB may be found at https://mariadb.com/kb/en/library/select-into-outfile/.

Prepare export folder such that it is accessible to everyone
Export data
Change access rules to export folder
Copy your data out of the export folder
Remove export folder

Please note that you will be copying your data to a new file, not moving it. This ensures you are the owner of the file, as opposed to the MariaDB server. Additionally note that your exported data will NOT include column headers/names!

Export Example

We will use the following to illustrate importing:
MariaDB username = jharvard
MariaDB database = jharvard_database
MariaDB database table = table_import
Import filename = SampleData.txt

Data:

COLUMN_1	COLUMN_2	COLUMN_3
25	Harvard	Way
86	Brattle	St

1. Prepare export folder such that it is accessible to everyone

Prepare an export folder and ensure it has the appropriate permissions:
mkdir /export/mdb_external/export/jharvard
chmod 777 /export/mdb_external/export/jharvard

2. Export data

Log into MariaDB:
mysql -h HOSTNAME -u jharvard -p

Within MariaDB, export the data:
select * from table_import into outfile ‘/export/mdb_external/export/jharvard/my_export.dat’ fields terminated by ‘|’ lines terminated by ‘\n’;

Please note our command has 3 sections:
(1) select * from table_import
specify what data you want export
(2) into outfile ‘/export/mdb_external/export/jharvard/my_export.dat’
specify the export file
(3) fields terminated by ‘|’lines terminated by ‘\n’
specify delimiters (the character which splits data or text into separate fields). More information about delimiters can be found here: https://mariadb.com/kb/en/delimiters/.

Official documentation for this command may be found at https://mariadb.com/kb/en/mariadb/select-into-outfile.