Quick Start

This guide will provide you with the basic information needed to get up and running on the HBS grid. If you'd like more detailed information, each section has a link to fuller documentation.

Step 1: Request an account

HBSGrid accounts are available to HBS faculty, research associates, and doctoral students. Please contact fill out the HBS Compute Grid Account Request Form to set up your account.

Temporary guest accounts on the HBSGrid are available for HBS faculty members' collaborators. These are granted on a 12-month basis and may be renewed as necessary. Collaborators should work with HBS Faculty and the RCS staff to submit the necessary documentation for the account request. Please see Logging In for details.

If you will be using compute resources on the cluster, please sign up for HBSGrid training once your account has been created. Our training materials are also available to view online for reference.

Last Updated 1/7/20


Step 2: Use VPN software when necessary

A VPN ensures that all communication between your computer and grid resources are encrypted, especially when using a public wireless network.

Connecting to the HBS VPN is necessary under the following conditions: 

  • Any off-campus connections
  • On-HBS-campus connections that are not through HBS Secure wireless or a wired ethernet

HBS-issued desktops and laptops have the necessary software pre-installed. Collaborators or individuals who need to re-install the software can see our VPN page for details. For a useful step-by-step list of instructions for connecting to the VPN, see the VPN Quick Reference Guide.

Last Updated 11/8/19

Step 3: Login via GUI or terminal

You can connect to the grid using a Graphical User Interfaces (GUIs) or by entering UNIX commands via terminal. 

NoMachine is a Graphical User Interface (GUI) client on the Grid. NoMachine allows users to utilize Grid software, RAM, and processor resources as though they were working on a desktop (i.e., Stata on the Grid looks like the desktop version of Stata). Unix/Linux knowledge is not needed.

For tips on customizing your NoMachine interface (e.g., changing screen size), please see our Quick Tips to NoMachine Interface Improvements.

Last Updated 11/8/19

For terminal (SSH) sessions to the Grid, initiate your SSH session with your Grid username to the hostname researchgrid.hbs.edu with the appropriate terminal client (e.g. 'ssh jharvard@hbsgrid.hbs.edu'). We recommend Terminal for Mac, and SecureCRT for Windows machines. Guest collaborators can download Putty to gain terminal access to the HBS grid.

See our Terminal Login to the HBS Grid page for more details on ways to connect to Grid resources. 

Last Updated 11/8/19

Step 4: Store and transfer your files

Once an account is created for you on the Grid, your home directory will be located at /export/home/<group>/<username> (e.g., /export/home/faculty/jharvard) to store things like your configuration files and custom scripts. Your home directory can only be accessed by you and have a storage quota, the size of which is determined by your group (e.g., doctoral, RA, or faculty). Please note that level 4 data cannot be stored on your home directory. 

If you are working on a project with one or more persons, a group project space can be created or may already exist for your work. The path for project spaces is /export/projects/<projectname>

We encourage the use of project spaces over home folders when appropriate for two reasons:

  1. these are accessible by all persons in the project (home folders are not); and
  2. the storage quota is much larger.

Please do not store desktop and laptop backups on the Grid storage, especially in project spaces. Please see Using Storage for more details.

You can transfer data to your home folder or project spaces via three methods:

A graphical tool like FileZilla is available cross-platform for Mac OSX, Linux, and Windows. Windows users who prefer SCP can download it from company's website.

Last Updated 11/8/19

Step 5: Determine what software to use

The HBSGrid offers several commonly used software packages including R, Stata, MATLAB, Mathematica, SAS, Stat/Transfer, and more. To get started using our pre-defined wrapper scripts, either select the software from the list of applications on NoMachine, or simply type the software name in your terminal window. Or you can use custom submission commands to run any piece of software that you'd like. 

If you require software packages that are not included among the default applications, you can access additional packages using our Software Modules feature, or use the instructions on our Installing Software on the Grid page to install needed applications.

For programming resources on the grid, you can refer to our Python and R package guide or our Software pages.

Last Updated 11/8/19

Step 6: Run a job or application

There are two different ways of doing work (running a job) on the HBSGrid:

  1. If you are logged in via the NoMachine GUI, you can run jobs interactively using packages in the Applications menu, or
  2. Open a terminal window to use the command line for more flexibility and to run several jobs at once. For terminal users, the command-line can be used to submit batch jobs and initiate interactive sessions using X-windows. To run jobs in terminal using Software Modules, please see our page on Accessing Software via Modules.

We ask that you choose your resources and memory footprint appropriately. Please close your sessions when you are not running jobs. Additional information is available in Running Jobs.

Last Updated 11/8/19

Step 7: Learn Etiquette

The HBSGrid cluster is a complex system of shared resources. Although we've taken great efforts to ensure that you can do your work in safe isolation, we expect users to follow certain guidelines to ensure fair access and to avoid interfering with other users' work.

Two important rules that will ensure fair access and use for everyone:

  • Avoid performing computations on the login nodes. Once you're in, start an interactive session (via drop-down menus or terminal commands) or submit a batch processing script or start an interactive session (see https://grid.rcs.hbs.org/interactive-vs-batch-jobs). We may kill any processes with high load (high memory use, high CPU use, etc.) that are running on the login nodes.
  • Be as accurate as possible when requesting memory or CPU cores for your jobs. Access to this shared resource depends on everyone honoring "Take what you need, and Need what you take." See Choosing Resources (https://grid.rcs.hbs.org/choosing-your-resources) for details. 

See the full list of Responsibilities and Acceptable Use (https://grid.rcs.hbs.org/rcs-acceptable-use-resources) guidelines.

Step 8: Get help and training

If you have any trouble with doing work on the cluster, first check the comprehensive Running Jobs pages (https://grid.rcs.hbs.org/running-jobs) and also our FAQs (https://grid.rcs.hbs.org/faqs). Don't forget to review our HBSGrid training materials (https://training.rcs.hbs.org/files/hbstraining/files/hbs_compute_grid_training.pdf), and use the search on this website.

If you still not have resolved the problem, please contact RCS (research@hbs.edu) for help. Help us help you by including  the job ID of the job in question, providing us with what happened, what you expected, and any error messages or screen shots of the problem. Log files and/or scripts will help as well.

For further research computing training opportunities, please visit our training website (https://training.rcs.hbs.org/).


Last Updated 5/6/20