The NSCL supports 2 methods of transferring experimental data back to users' home institution: Network file transfer and LTO/DLT tapes. Some users have chosen to take their data home using various unsupported options. Two of those are described here.
If you use a supported method to transfer data to your home institution, the NSCL computer group will assist you if you have problems. This assistance will be done within the normal NSCL computer group problem response committments. Therefore, please plan ahead rather than attempting to write your data at the last minute.
If you choose to use an unsupported method to transfer your data to your home institution, our ability to help you in case of problems is limited and will depend on the other support committments the computer group has at the time.
![]() | Write Tapes Too!!! |
|---|---|
We would like to emphasize that a BACKUP of the original event data should be created on a reliable medium such as LTO/DLT tape. The options discussed below should not be considered as a means to backup data. The previous chapter describes the recommended and supported ways to create this reliable backup. |
Taking your data home on DLT or LTO tape is a supported option.
Writing your data to tape as described in Section 2 is one supported mechanism for transporting your data home. If you write tapes for transport home, you are encouraged to write another set for storage at the NSCL.
The NSCL will store tapes of data from
your experiment for you. NSCL data tapes are stored in an
environmentally controlled room. If you require a copy of tapes
stored at the NSCL, contact <helpme@nscl.msu.edu>.
Transferring your data home via the network using ssh/sftp is a supported option.
First transfer your data to the evtdata areas at the NSCL as described in Section 2, as the experimental systems are firewalled in both directions from systems external to the NSCL. File the tapes you created and validated for storage at the NSCL.
If you have sufficient time before you leave, you can push the data from your evt areas to your home institution. If not, once you return you can pull the data across the network as long as your account at the NSCL is alive (2 weeks from completion of the experiment).
Suppose you have been allocated /evtdata/myexperiment and transferred your data tapes into that directory. To push the data to the directory somedir on the account myaccount on host myhost.myinstitution.edu; Log into your nscl account on a spcie system and:
bash$ cd /evtdata/myexperiment
bash$ scp -r * myaccount@myhost.myinstitution.edu:somedir
When prompted accept the host key offered and type in the password of
your account on the remote system when prompted.
To pull the same data, login to some system at your home institution and:
bash$ cd somedir
bash$ scp -r nsclaccount@nsclgw1.nscl.msu.edu:/evtdata/myexperiment .
where nsclaccount is your nscl experiment
account. Once more accept the offered host key and type the password
of your NSCL account when prompted.
Using USB external disks to transport your data to your home institution is not supported at this time. If you run into problems with this option, please try a supported option.
USB external disks generally work seamlessly on the Windows PCs in the Data-Us. There are USB ports in the Windows PC's in the Data-U Simply plugging an USB device into a windows PC will automatically mount it, usually as the D: or E: drive. Both Buslink and Lacie disks are known to work, but others will likely work as well.
We've seen transfer speeds of up to 9GB/hour using this method. You should therefore plan ahead so that you have sufficient time to finish copying your data to disk before you leave.
If you choose to use this method, The USB device must not be just unplugged when the transfer is finished. Instead the Windows option to 'Safely Disconnect'. External Hardware' should be used to dismount the device. Otherwise, your data will be corrupted.
Transporting your data home via CD or DVD optical media is not a supported option at this time. If you run into problems with this option, try one of the supported options.
Writing CDs or DVDs is another option. If your experiment has taken a lot of data, this is not a good option because the media is relatively small compared to both tape and disk (700Mbytes per CD, and a few GB per DVD depending on the format). Furthermore, DVD media comes in many flavors and not all media are readable on all DVD drives.
There is a publicly available CD/DVD writer on the 'tapehost' machine in Data-U2. While not supported by the NSCL many users have used it successfully. We do know of least one failure.