System Administration

The Molecular Spectroscopy team of NASA/GSFC's Laboratory for Extraterrestrial Physics relies on a Sun compute/file server for its analytical work. Data collected on field runs over the last several years is stored on this server. The value of this data is inestimable, because much of it relates to objects of opportunity (i.e., comets Hyakutake and Hale-Bopp) and is irreplaceable. Furthermore, the team generates "ancestor" data from the raw data which amounts to many hundreds of megabytes per year. This information represents unique scientific effort and thousands of man-hours.

Because this data was in disarray, I designed an organizational protocol for old and future information. This scheme distributed the data efficiently among the server's disks and allowed for intuitive human searching and easier backups.

To ensure data security, I instituted a two-backup system: an on-site backup and an off-site backup in case of physical mishap such as fire. These backups are performed periodically. I also wrote software to automate all backup procedures and trained members of the team in its operation. The software scans the file system and makes backups as needed on a daily, weekly, and monthly basis.

Because the group's daily activities revolve around the use of IDL and IRAF software, I maintained these systems along with the custom libraries used for analysis. This involved installing, patching, and compiling various facilities in a variety of languages including C, Fortran, and IDL.

The group requires FTP and HTTP service, so I constructed and maintained servers for these functions also.

When hardware failed or became obsolete, I repaired or replaced it, performing all research for new technologies and ushering purchases through the NASA bureaucracy.

Security is a problem at a "target" institution such as NASA, so I took many precautions including secure shell installation, TCP wrappers, tripwires, updated patches, service limitations, and a number of custom modifications. The group's facilities consist of three UNIX machines, seven Apple computers, and assorted PCs running Windows and Linux.

All of the above activities, including hardware identification, network information, help contact information, and detailed instructions for re-creating any part of our systems was summarized in a 40-page booklet which I wrote and distributed to the members of the team during a training session. This further insured data integrity and reliability by doing away with a situation in which one critical person carried all important system knowledge.