Ex Libris MELvyl Tells (EL Mel Tells)!
Volume 1, Number 4, October 2002
In this issue:
Command Searching
A new feature of the web version of the Melvyl-T catalog is the Command Search option. Command Search, utilizing the Common Command Language (CCL), is among the most powerful forms of searching within Melvyl-T because it allows you to combine more than one of the optional limits in a single search. Although CCL can be used for simple searches, it is an especially effective way to perform complex searches, as it affords you a great deal of control over your search results.
The essential building blocks of a Command Search are the three-character index code (upper or lower case) followed by one or more search terms. Command Search accommodates truncation and Boolean connectors.
For example, if you would like to find out whether UC San Diego owns the Henry Reeve 1840 English translation of Democracy in America by Alexis de Tocqueville, your Command Search might be:
(aut=tocque*) and (wau=reeve) and (wti=democracy america) and (wyr=1840) and (wln=eng) and (wid=ucsd)
By default, unless you enter a three-character index code, Melvyl-T will automatically insert the index code “WRD” and perform a keyword search. For example, the following three searches are equivalent and will retrieve identical results:
(mesopotamia or babylonia)
(WRD=mesopotamia or WRD=babylonia)
(wrd=mesopotamia or babylonia)
Command Search is at present the only place where you can select more than one of the Optional limits in a single search and use Boolean operators. For example, to see if the de Tocqueville book above is available at both UCSD and UCLA, your search would be:
(aut=tocque*) and (wau=reeve) and (wti=democracy america) and (wyr=1840) and (wln=eng) and (wid=ucsd or ucla)
Or, assume a comparative literature student is looking for versions of Shakespeare’s Hamlet in French or Spanish. The search would look like this:
wti=hamlet and wln=fre or spa
The three-character index codes used for Command Search in the Web-based version of Melvyl-T are generated by the system software and cannot be modified. The index codes used in the Telnet version of Melvyl-T will remain the same as those used for legacy Telnet Melvyl.
A complete list of CCL three-character index codes may be found within the
Melvyl-T HELP environment.
http://mel-t.cdlib.org/F/?func=file&file_name=help-search-command-cdl90
Campus Snapshots, or How to Build an Enormous Database
Interviewing Lynne Cameron of the CDL Content Group for this article gave me even greater respect for my CDL colleagues than ever before. And even though I’ve been hearing about the snapshots for some months, it is only now that I understand the complexity involved in accomplishing this feat. Months in the planning, the loading and indexing of some 23-million campus records took two months.
At the beginning of the building of Melvyl-T, campuses were given the choice of having their records unloaded from legacy Melvyl, then loaded into the new database, or of creating new campus snapshots. All of the campuses chose this latter approach, since it provided them the opportunity to cleanup their databases and start fresh. (Some of the other input streams into Melvyl-T will take the unloading and reloading approach.)
Historically, all input files for legacy Melvyl (CAT and PE) were FTPed to the mainframe. Before receiving the campus snapshots, CDL staff first converted all regular input file ingestion from the mainframe to a UNIX environment in preparation for Melvyl-T. (The set-up for this process, which included the creation of an FTP Virtual Host, was done by the very able Debra Bartling and assisted by Raymund Ramos, in coordination with staff from the Data Center.) Because the input files would continue to be run into legacy Melvyl, the mainframe processing streams were modified to pull the input files from the FTP Virtual Host. For each source, CDL staff compared the output of the legacy streams using files that had been FTPed to both the mainframe and FTP Virtual Host. The records were tested and retested for accuracy, comparing the outputs from both to ensure they were the same. An enormous amount of testing took place by Randy Lai and Virginia Sinclair, with Virginia keeping in touch with the campuses during the process. Although up to three sources were migrated simultaneously in a single week, the process of migrating sources to the FTP Virtual Host took two months. And all the while, legacy Melvyl is being maintained, with weekly updates loaded to both versions of the catalog.
This project was a massive one, requiring cooperation and coordination of CDL, Data Center, and campus staff. Sending campus snapshots to the CDL for Melvyl-T is not like sending snapshots to your parents or grandparents or sisters or brothers. For some campuses, it took several weeks to send the records to us, in chunks of 50,000 records at a time, ten files per day! The files were FTPed to the Virtual Host on UNIX cluster machines Everest and Olympus.
Since the production AIX machines, Hermes, Mercury, and “little” Athena, were not ready to accept the snapshots when the process started, the records went to the Tivoli Storage Management (TSM) Archive from the FTP Virtual Host. (CDL was the first at UCOP to use this process.) Consultant Bob Read (aka Waffle Dude, since he occasionally spontaneously made waffles for all CDL staff and passersby) wrote an important program that was a Virtual Host monitor—it detected when the records arrived, and pushed them into the archive when the files were complete. Records remained in the archive until Mercury, the production machine, and Hermes, the failover machine, were ready. Randy pulled records out of the archive and staged them for loading. At this point, they were run through CDL converters and fast loaded. Rebecca Doherty, an amazingly detail-oriented bibliographic analyst, reviewed the statistics for 500 different files. Changes were made to the converters along the way, sources were re-run, and the files were at last ready to go into the Aleph software.
While all this was going on, CDL Content Group staff worked quietly and cohesively, with great cooperation and in partnership with the Data Center and campuses. I mentioned that the very same people who did all this work are also keeping legacy Melvyl running, and will continue to keep the two versions running in parallel through June. But what I neglected to say is that they these same people are also keeping the CDL-hosted A & I journal article databases running through the end of 2002. Then they’ll be busy decommissioning these databases! Kudos to all the CDL staff who made this very complex, intricate process fall into place, and seem as simple as…well, as taking a snapshot.
CDL Staff Profiles
Continuing on with profiles of those who are helping to build Melvyl-T, meet
the following people:
| Juanita Jones | |
 |
Juanita provides administrative support for several of the people involved
in different aspects of Melvyl-T including John Ober, Rosalie Lack, and
Ellen Meltzer, as well as others in CDL. She is one of the key persons
who arranges special events, conference calls, travel, catering, and those
ever-important reimbursements! She tries to keep everyone sane.
Juanita also marks-up El Mel Tells, making it available to us all on the Web. She has a hidden talent as a photographer, takes all the photos in El Mel and doesn’t bat an eye when someone wants a picture taken more than once.
|
Raised on a farm near Bogalusa, Louisiana, Juanita has a son, Brandon, now 20. The summer of 2003, Juanita's family reunion in Louisiana will number over 300 Joneses! She also answers to the names of Nita, Dot or Dottie. Ask her why! |
|
Rosalie plays a critical role in the public end of the Melvyl transition--from creating and maintaining public Web pages, to editing CDLINFO (where transition articles are easily found--they’re always item #1), to serving on the Melvyl Education/Usability Transition Team, to creating the campus Usability Toolkit. Rosalie has played a vital role in creating wonderful materials for the campuses to use when conducting usability tests, including testing the first two versions of the Melvyl-T interface. |
Rosalie Lack |
She’s grateful to the campus Evaluation Liaisons for conducting the tests, and did an amazing job of analyzing the results received from them so that we were able to make necessary improvements in the user interface. She acts as a consultant to the campuses on a regular basis, as well, on issues of usability and other surveying and testing techniques. (Working on the Melvyl-T project is just a small part of Rosalie’s responsibilities at the CDL.)When the fog rolls into San Francisco on those cold summer days, Rosalie longs for her native Jamaica, where she lived until the age of 11. |
| Robin Davis-White | |
 |
Robin provides administrative support services for the majority of the
people at the CDL involved in the Melvyl-T project. She does some Web
work, sets up meetings, conference calls, and makes travel arrangements
for these staff members.
She is also working on a degree in information technology from the University of Phoenix, and has graduated from Academy of Web Design and is taking video classes locally. Robin never stops--she has an incredible array of interests and hobbies that she does when she's not at work or school, ranging |
from video and movie-production, making stained glass objects (including a door), and gardening and landscaping (she has built three ponds with lily pads and a waterfall in her back yard). She also writes poetry and music--you name it and this talented CDLer does it! |
|
John has had an intimate and changing role with the Melvyl-T Project. First, during the period in which Beverlee French was Interim Director of the CDL, John, Bev, and Laine Farley were responsible for the general oversight of the project, including hiring people to manage its various aspects. Since May 13, 2002, however, he has taken on the more hands-on role as Interim Director of Technology at the CDL, in addition to his duties as Director, Education and Strategic Management. |
John Ober |
How does he do it? He helped organize the Technical Leads Team, which meets on a
regular basis to keep all the technical threads of the project together.
He participates in weekly phone calls with CDL and Ex Libris staff.
He attends regular meetings with the Melvyl Project staff and Melvyl Transition Team leaders, and on a regular basis brings background issues for discussion and advice to SOPAG. He has also been closely involved, with Raymund Ramos, in Performance Testing for the new catalog. John has primary responsibility for the CDL's contributions to the system architecture discussion and decisions, including backup and recovery of Melvyl-T (which is stored within 5.8 terabytes of new disk purchased for the system). he other major player in this discussion is the UCOP Data Center. The Data Center plays a critical role in managing the hardware and software that hosts Melvyl and its servers, Pinatubo, Hermes, Mercury, and Athena. System architecture planning involves a complicated and complex six-way partnership including CDL, the Data Center, which manages the Tivoli backup software, Ex Libris, IBM (and their consultants), and Oracle. Find John juggling among these partners! Two unrelated things in John's past have helped him in his current job (the answer to the above question: how does he do it?): a 43-day, 8 hours and some minutes bike ride across the country when he was 21, and his 9 months in Benin as an ALA Book Fellow helping the Benin National University with library automation planning (with their 3 and a half functioning microcomputers). When he thinks something is not possible, John draws on these experiences to tell him whatever it is, it can be done! |
| Mark Reyes | |
 |
Mark is a relatively new CDL employee, so when I asked him what he does
on the Melvyl-T project, he retorted, “Good question!”
Mark, who works in the Technology Content Group, does not come from a library background, so the world of bibliographic records is new to him. He brings skills in programming and in the production environment from his previous work at the Jet Propulsion Laboratory in Pasadena, and a degree in computer science. |
Mark is on the team that has been loading the campus snapshots
containing the 24 million or so Melvyl records into production Melvyl,
and verifying that the records are correctly loaded. Since this is the
first time this loading is taking place in the new environment, some of
this checking is done manually.
Mark has also been involved in setting up the IBM Tivoli job scheduler software that controls the job stream from production, to load, to returning errors to the campuses. He has set the job scheduler up for the development machine, and the whole production cluster. Mark is a cycling fanatic, and he BARTs and cycles into work. If Mark were not working at the CDL his ideal job would be as a bike courier! |
|
Michael is a systems administrator for Melvyl-T. Some of the things he does in this role are to move files around, configure the Aleph software for new instances (copies) of the database, and write scripts that ensure that Oracle, Apache, and Aleph are all up and running--and if not, to tell us so. He's also created a script so that there is a nightly clean-up of the VIR01 table, a pesky table that fills up frequently and causes all kinds of trouble when it's filled. |
Michael Russell |
Michael did sys admin work at UC Davis for ten years, and originally
learned programming in a FORTRAN class
at the University of Alaska.
Michael works at home on Tuesdays, so he conference-calls in to biweekly Melvyl Project Meetings. At these times, we're lucky enough to speak not only to Michael, but to his Jack Russell terriers Millie and Sherwood, who participate with an occasional background bark. (When Michael bought his house, Sherwood came with it--go figure!) Michael is also the force behind weekly Wednesday CDL Survivor lunches, in which all CDL staff are invited to come, a great chance for people who work on different parts of the floor to meet and talk. Michael is an amateur violist who played in a chamber group in Sacramento which tended to be focused on performing, but Michael took up the instrument with the idea of being able to get together with friends on Friday nights to saw out a few string quartets, but, more importantly, to partake of wine, cheese, and bread! |
Got a burning question? Tell us what topics you want to see covered here. CDL Feedback