It is important to understand that Globus transfers do not operate in the same way as cloud storage services. Such services require you to upload your data to the cloud and then "sync" on the destination computer to pull the data down from the cloud. Instead, Globus transfers are performed directly between two systems that are set up to provide Globus "connections".

Connections

Connections are named locations in the Globus network that provide access to data. Collections may be supplied by computers that are running Globus software or through Globus "connectors" that provide access to cloud storage. When you perform a Globus data transfer, you only need to know the names of the two collections between which the data will be copied, and not the ID or location of the underlying computer systems that provide the collections.

Endpoints

Some Globus collections are provided by Globus Connect software that is running on a server or a personal computer. Such software is said to create an endpoint that can support one or more Globus collections. The Globus Connect software registers its endpoint with the Globus network, which then allow transfers to be made to and from the endpoint's collections.

Institutional computers (like TACC resources) are already running Globus Connect software, and you do not have to do anything in order to use their endpoints in data transfers. If your institution is set up for Globus file transfers, it will have endpoints that you can use when transferring data. However, if you want to have an endpoint on your personal computer, you must run the Globus Connect Personal software on that computer. This advanced topic is covered in Globus Data Transfer - Advanced.

Accounts and Privileges

A user must be logged in to Globus in order to perform a data transfer. The Globus Accounts page describes the different types of accounts that are available and how to set them up. Users must also have appropriate read and write privileges for the endpoints used in a transfer. These privileges can be established by providing credentials when specifying the endpoints or by using an account that grants the privileges for those endpoints.

 
©   Cornell University  |  Center for Advanced Computing  |  Copyright Statement  |  Inclusivity Statement