LMC and MMG Update

15 August 2001

UPDATE

Hardware:

-Test server: The test server which is the Elsevier project server for which we bought additional disk, has installed on it these new Orbis2 disk drives and an operating system. Wesley Most and Richard Crane completed this work before the end of July.

In order to have a Yale database for the SUN certification test quickly installed Oracle and Voyager software on the test server. The installation of Oracle and Voyager will be repeated onsite at Yale and will include technical training for Yale staff. Endeavor has yet to set a date for this installation/training.

In the meantime, with help from the Implementation Team, Wes is conducting some system load tests on the server.

-SUN certification test: During the week of August 20th Yale (Audrey and David) and Endeavor will conduct a certification test at the SUN testing center in California. For this test, Gail Barnett extracted all Yale bibliographic (5M) and authority (1.5M) records. These were ftp'd to the test sever where Endeavor converted them to Voyager bib and auth records.

This data was then copied to tape and shipped to CA. (Approx. $4000 was spent from the Orbis2 budget for the tape backup unit.)

During the week of Aug 20th we will test the new servers, Instant Image (which we want to use to keep redundant systems in sync) and a fix to the SUN storage devices we want to use. We will conduct 2 benchmarking tests - a full database index rebuild and bulk import of 20,000 bibliographic records. We will also test the Voyager clients with some basic scripts to ensure Voyager runs on the new servers.

Based on the test results, ITS will recommend a production server configuration to the MMG.

Recently paid Endeavor invoices:

-40% of the software license fee that was due upon contract signing

-$250 for our share of the escrow account

Endeavor Unicode Task Force:

Audrey went to Endeavor on 8/13-14 to attend the first Endeavor Unicode task force. This task force will participate in the development of Voyager support for non-Roman scripts. The next phase in this project is the conversion of the Endeavor MARC databases from their current character set (Voyager RLIN) to UTF-8. Additionally, the Cataloging client will be modified to enable Unicode character input. Modification of the Acq and Circ clients, culturally appropriate indexing of the 880s and support for bi-directional scripts are all scheduled for a later phase.

The task force reviewed issues such as:

-Forward mapping to UTF-8 for the database and incoming records (OCLC, RLIN, etc.)

-Reverse mapping from UTF-8 for exports to RLIN, OCLC, auth control vendors, table-of-contents vendors, etc.

-Normalization of Unicode characters for searching and indexing (but not culturally appropriate indexing of 880s)

-Unicode character input and display methods.

-Z39.50, ASCII OPAC, and WebVoyager (display, print, download, and email implications).

The most difficult issues concern forward and reverse mappings especially of the pre-composed and decomposed characters (that is, a character + diacritical mark that is combined, or a character + diacritic as separater characters) and normalization.

This project complicates Yale's implementation and it suggests a significant delay in the availability of full-support for non-Roman scripts:

- The glyph server (Phase I support for non-Roman characters) is not working. In its current form it displays only Chinese characters and it frequently displays the wrong character. Sites with larger CJK collections (Hawaii, Princeton, Pitt) cannot put the glyph server into production. As a result of the TF meeting, Endeavor will focus attention on the glyph server. Mapping for display is similar to mapping for data conversion but much safer. If the display mapping is not working after a year of development, we should anticipate problems in forward and reverse data conversion.

Note that:

- Development of Phase II is just starting. The goal is to have the data conversion in beta testing by Jan and in production in the Spring. This schedule means that in order to go into production with a UTF-8 database we will need to participate in the beta testing phase. During the final production loads, we will need to immediately follow the conversion of our data from NOTIS with a conversion to UTF-8. This timing adds a significant level of complexity to our implementation. It also increases the risk of loss and corruption of data.

- Given that Phase II will not be completed before the Spring it is unlikely that support for bi-directional scripts will be in place by the end of 2002.

The next phase after conversion to UTF-8 will be indexing of headings in vernacular scripts (the 880s).

- Prior to the production implementation we will have to upgrade from Oracle 8 to Orcale 9 (an additional complication).

- All Tech services workstations will need Windows 2000.

SFX Implementation:

On 8/10 Ex Libris replied to Yale's license agreement modifications with a new agreement. In general they accepted our changes. These include: agreement to update the Knowledge base every 3 months; ability for Yale to >>cancel the Knowledge base maintenance if the file is not updated every 3 months; an implementation timeline that gives Yale 2 months in which to try to implement SFX - if we fail, we can return it and will have to pay for services (training), only. The license agreement is being reviewed by Lib Business Office, University Purchasing and Kim Parker (SFX Project Manager) before it is forwarded to Alice for review this week.

Ex Libris is setting up a proof-of-concept test of links back into our NOTIS Orbis webpac on their test SFX server.

Kim Parker is working with Ex Libris to schedule SFX training. Training will take place at Yale during the first week of Sept. Exact dates are still to be scheduled.

Richard Crane is investigating the feasability of a Linux SFX server rather than a SUN server. A Linux server will cost about 1/3rd as much as a SUN server.

Implementation Team:

-The Data Migration work group is actively generating specifications for Orbis clean-up projects. To date Fixed fields for annuals and old Geac missing/BFR codes have been corrected.

See: http://www.library.yale.edu/orbis2/implementation/DataMigration.htm

for complete list of clean-up projects.

-The Systems group is identifying local programming requirements for the new system and development timeframes for each.

-The Acq/Serials, Cat/auth control, Circ and OPAC groups finished Voyager functional training last week. Each group is working through their part of the System Administration tables. The Locations subgroup is addressing the huge task of mapping NOTIS locations codes to Voyager codes.

Circulation rules a defined differently in Voyager and as a result we must generate this mapping table. More specifically, location is not a variable in the Voyager circ policy matrix. Additionally, Endeavor and other former NOTIS libraries have advised us that we should reduce the number of location codes as much as is possible.

Each member of the location group is looking at their own library's location codes and determining which codes must be retained and why. From this examination the group will create a list for reasons for the retention of a location code. They will then review location codes with the owning libraries in order to construct the NOTIS-to-Voyager location mapping table.

Next week an additional subgroup will be formed to address our indexing requirements.

-Kate Reynolds is organizing time management and effective committee management training sessions for the Imp team members.

Project Plan:

Endeavor has yet to draft a detailed YUL Project Plan. We do not know with whom and when we will start working on authentication, all of our data migration special requirements, correct handling of Old Yale call numbers, course reserve modifications, etc. In response to our initial inquiry they determined that they would come to Yale for a Project Planning meeting on Aug 27/18. They will meet with Audrey and the Chairs of the Imp Team groups.

Audrey discussed the problems associated with not having a Project Plan with our Endeavor Project Manager during the Unicode TF meetings. We now have contact names in development and Endeavor has a better sense of our need for an implementation schedule. Endeavor will provide a draft Project Plan next week. This plan will be amended by Audrey to include Yale-only implementation steps. The entire plan will be reviewed in detail Aug 27/28.