Stanford University Libraries (SUL) has completed the first five months of its migration from mainframe NOTIS to Sirsi's Unicorn. Included in the migration are the Crown Law Library, the Hoover Institution library, and the SUL/AIR libraries: Green Library, Meyer Library, seven science branches, and three humanities and social science branches.
The decision to migrate from NOTIS to Unicorn (version 8 beta) was made in February, 1996 and put on a fast track for implementation by September 3 (circulation and cataloging) and September 10 (acquisitions). All NOTIS transaction activity ceased on August 31. Serials check-in was and is still manual. Authorities were not in NOTIS, but a file of records from a complete vendor pass was integrated with the Unicorn bibliographic dataload.
Socrates, the mainframe OPAC developed at Stanford in the early 1980's, had been updated nightly from NOTIS bibliographic and holdings transactions and now is fed nightly from Unicorn. Circulation status is delivered from Unicorn to Socrates upon explicit patron request for status, as was the case with NOTIS. Socrates II, the web replacement for Socrates, is in preliminary release for the Stanford community. Using the Sirsi search engine and many WebCat features and tools, Socrates II is targeted for general release within March and for full functional release this summer. The original Socrates application eventually will be retired from service.
SUL's significant business need to migrate to the Unicorn transaction system before fall quarter 1996 compelled us to make best guesses for conversion, indexing, and system parameters. By and large we are pleased with the quality of our decisions and are impressed with Sirsi's first rate technical assistance throughout the conversion. We currently are in production with version 8.1.4.
Now that experience, the best teacher, has taught us much about the technical and functional infrastructures of Unicorn, we are moving to the phase of implementation during which Sirsi's APIs (Application Programming Interfaces) and other Sirsi tools will serve us as we tune daily transaction functionality to our complex local needs. Local proof of concept work has caused us to be quite optimistic about our ability to use the Sirsi tools to streamline online transaction work flows and to apply the tools to Stanford-specific functionality in technical processing, including the Stanford model for Fast Track processing, and circulation.
We were correct when we anticipated an intense year of implementation challenges and surprises; we expect 1997 to be as intense but with fewer bumps in the road. Heroic staff efforts at every level, in every library unit, are focused on keeping the work moving, analyzing problems, postulating solutions, and documenting resolutions. And after all the stress, hard work, and unforeseen events that are a part of converting a large library's processing to a new system, we can affirm that Sirsi's Unicorn does in fact deliver what we were looking for.
The following information reflects our accomplishments with the conversion and some short-term deliverables that fit into what we consider this first phase of implementation:
Circulation
- inverted and loaded 40,000 patron records and 65,000 circulation transactions
- built link between University faculty/student/staff records for nightly load of patrons
- released web-based Reserve List search function (unplugged home-grown application)
Bibliographic Database And Cataloging
- converted, loaded, and indexed 3.2 million bibliographic records and their related holdings records and 3.4 million item records (see below for indexing times)
- integrated 1.2 million authority records from vendor pass
- installed nightly link to Socrates for bibliographic and holdings updates
- installed SmartPort for Z39.50 bibliographic import from RLIN, OCLC, and LC Acquisitions
- converted and loaded 55,000 active orders
- built data link from Unicorn paid invoices to University accounting systems
- [Q2 `97] develop EDI (EDIFACT) links to US (Yankee Book Peddler) and a European (Harrassowitz) book vendors [Q2-3 `97] apply Unicorn API technology to Stanford Fast Track monograph receiving
Serials
- agreed with Sirsi to provide detailed analysis of Unicorn Serials check-in for complex serials environments [deliver analysis Q2 `97]
- [Q1 `98] begin conversion from manual check in
OPAC (Socrates)
- built nightly data loading from Unicorn to Socrates database
- installed pre-release version of Socrates II client, a web interface to the Unicorn database [public release Q2 `97]. Authority references and instant display of circulation status are integral to Socrates II and will be new to our community along with many Unicorn features, e.g., adjacency searching, sorted results.
System Components And Operation Indexes
- headings index for 3.2 million MARC records: 5 days *
- union keyword index of full MARC record: 4.3 days
- keyword index for largest single library: 1.3 days
Server
- 6 CPUs (PowerPC 604) with 2 GBytes of memory
- Raid-1 disc (Unicorn @ SU is 40 GBytes x 2 for mirroring)
Backups
- 100% mirror for Unicorn's programs and files
- hourly magnetic archive copy of logs of update transactions
- weekly snapshot to separate disk media and off-site tape image
Downtime (first 5 months in production)
- one hardware crash caused by a UNIX operating system bug
- zero Unicorn software crashes
- downtime during backup: 2 hrs of weekly full snapshot during file copy verification (maintenance schedule for University production systems: Thu. 5:00 - 7:00 a.m.; Sat. & Sun. 5:00 - 8:00 a.m.)
