Globus Online Data Transfer

Globus is the preferred and most efficient way to transfer data between DTNs and external storage systems. To use this service, you must have a NOAA login name and a working RSA SecureID token. You can invoke Globus functions either through a web interface or from a command line interface (CLI).

Click here to access Globus Documentation. Click here to review the Globus Tutorial

Overview

An endpoint is a file transfer location (computer/server) accessible to Globus. A collection is a server with a related access method to files. Untrusted collections can transfer data to and from anywhere. Trusted collections can transfer data to and from other vetted collections. When you log into Globus and click Collections, you can see what collections are shared with you, and also those that you share with others. Globus lets you navigate through collections to find source and target endpoints for your transfer, then select directories or files to be transferred. The transfer itself is a background process.

To copy a file, several files, or an entire directory between two systems, navigate to Globus.

Locate the source and target endpoints by their given names and follow these steps:

1. Authenticate yourself to both endpoints.

2. Select the Directory Listing panel for each Endpoint.

3. Pick a directory in each panel for your source and destination.

4. Click START to initiate the transfer.

Example

1. Navigate to globus.org.

2. Select “existing organizational login” NOAA RDHPCS. The File Manager page displays.

3. Select Collection, and choose the file system “noaardhpcs#niagara_untrusted”. If necessary, authenticate with username and RSA password.

4. In the File Manager, select Path: /collab1/data_untrusted/anonymous/from Orion

5. Repeat for the other endpoint: msuhpc2#Orion-dtn

6. Select files and directories, and click Start.

RDHPCS Globus Collection Summary

Globus Connect Service is available on the following RDHPCS and partner clusters.

Analysis / PPAN

Cluster	Display Name	File Systems	Site	Access
PPAN	noaardhpcs#ppan	/archive, /home, /nbhome, /work, /xtmp	GFDL	Trusted hosts
PPAN	noaardhpcs#ppan_untrusted	/collab1/data_untrusted	GFDL	Anywhere

Gaea

Cluster	Display Name	File Systems	Site	Access
Gaea	noaardhpcs#gaea	/gpfs/f[56], $HOME	NCRC	Anywhere

Hera

Cluster	Display Name	File Systems	Site	Access
Hera	noaardhpcs#hera	/scratch1, /scratch2	NESCC	Trusted hosts
Hera	noaardhpcs#hera_untrusted	/scratch1/data_untrusted, /scratch2/data_untrusted	NESCC	Anywhere

Jet

Cluster	Display Name	File Systems	Site	Access
Jet	noaardhpcs#jet	/mnt/lfs4, /mnt/lfs5	NESCC	Trusted hosts
Jet	noaardhpcs#jet_untrusted	/mnt/lfs4/data_untrusted, /mnt/lfs5/data_untrusted	NESCC	Anywhere

Niagara

Cluster	Display Name	File Systems	Site	Access
Niagara	noaardhpcs#niagara	/collab1/data	NESCC	Trusted hosts
Niagara	noaardhpcs#niagara_untrusted	/mnt/lfs1/data_untrusted	NESCC	Anywhere

Orion

Cluster	Display Name	File Systems	Site	Access
orion	msuhpc2#orion-dtn	/work, /work2	Orion DTN at MSU	Anywhere

Hercules

Cluster	Display Name	File Systems	Site	Access
Hercules	msuhpc2#hercules	/work, /work2	Hercules DTN at MSU	Anywhere

NOAA RDHPCS Globus Endpoint Types

Note

It is preferable to use Trusted Endpoints for data transfer.

NOAA RDHPCS Globus Endpoints are either ‘’trusted’’ or ‘’untrusted’’.

• All RDHPCS systems provide DTN’s

• DTNs have full access to the back-end file systems.

• DTNs only accept connections from pre-authorized sites. If your site can’t access the DTNs and you need that capability, submit a help desk ticket. If the security team approves, your site will be pre-authorized.

NOAA RDHPCS UDTN’s (Globus Untrusted Endpoint)

UDTNs can accept connections and transfer data to and from any location. UDTNs have access to a specific directory of the back-end file system, where files can be staged solely for the purpose of transferring data.

Since your project space is not accessible from the UTDN, transferring data to and from RDHPCS systems using the UDTN’s is a two-step process.

1. Copy the data out of your project space to the staging area and then pull data out of the UDTN from the remote machine.

2. To transfer data back to the RDHPCS system, push the data to the UDTN, then copy the file(s) from the staging area to your project space.

NOAA RDHPCS Object Stores in the Cloud

RDHPCS maintains Cloud Stores in Microsoft Azure, Amazon S3, and Google Cloud. From the Globus perspective, connecting to these types of resources is identical to any other endpoints serving DTNs.

The RDHPCS Globus plan offers connectors to access data to and from a public site available via AWS resources.

1. Navigate to globus.org.

2. Select “existing organizational login” NOAA RDHPCS. The File Manager page displays.

3. Select Collection, and search for NOAARDHPCS# collections.

4. Once you can see the file lists, you can use the “File Manager” to move the files between the desired endpoints.

Globus Command Line Interface (CLI)

The CLI is available on Jet, Hera, and Niagara.

If you would like to use Globus-cli, either on your personal machine or on a system where globus-cli is not installed, you can install it easily . Instructions to install and use the Globus CLI are available in the Globus documentation CLI section.

Transferring Data to and from Your Computer

To transfer data from your laptop/workstation to a NOAA RDHPCS system, you can

• use scp to a NOAA RDHPCS DTN (using pre-configured ssh port tunnels)

• use scp to a NOAA RDHPCS UDTN

• use Globus Connect Personal to transfer data between a NOAA RDHPCS UDTN and your local laptop/workstation.

Note

NOAA RDHPCS considers your laptop/workstation a Globus Untrusted Endpoint.

Some benefits of using Globus Connect Personal with UDTNs:

• Data can be transferred directly between your computer and an Untrusted Endpoint.

• Faster transfer rates as compared to scp and sftp.

• Data transfers automatically suspends and resumes as your computer goes to sleep, wakes up, or reboots.

• The mechanism for transferring data between your laptop/workstation (Untrusted Endpoint) and a NOAA RDHPCS UDTN is exactly the same.

Please see Globus Connect Personal for information about setting up your laptop/workstation as a Globus Personal Endpoint.

Warning

Please note the following warnings when using the Globus Online transfers.

• Globus transfers do not preserve file permissions. Arriving files will have (rw-r-r-) permissions, meaning arriving files will have user read and write permissions and group and world read permissions. Note that the arriving files will not have any execute permissions, so you will need to use chmod to reset execute permissions before running a Globus-transferred executable.

• Globus will overwrite files at the destination with identically named source files. This is done without warning.

• Globus has restriction of 8 active transfers across all the users. Each user has a limit of 3 active transfers, so it is required to transfer a lot of data on each transfer than less data across many transfers.

• If a folder is constituted with mixed files including thousands of small files (less than 1MB each one), it would be better to tar the smallfiles. Otherwise, if the files are larger, Globus will handle them.