Drawing Piano Keys on the Table

Lessons Learned in Migration Planning

Dreamt I had drawn piano keys
on my kitchen table. I played on them, mute.
The neighbors came over to listen.
Sorrow Gondola No. 2
By Tomas Tranströmer

When preparing to migrate library systems, time is the ultimate luxury. The professional literature is awash in articles highlighting the importance of taking time pre-migration to thoroughly review library system data, comb through item status reports, and reconsider policies, workflows, and matrices; to clean (and delete) as much data as possible; and to consider a physical inventory. This excellent advice guided the project timeline of the last system migration at Wright Library, Princeton Theological Seminary in 2019-2020. But next time, we’ll do things differently.

Drawing Piano Keys on the Table

Our context was not an uncommon one for that era. The Library was on the Ex Libris Voyager ILS for over twenty years, from 1999 to 2020. Knowing that migration from a long-term legacy system would be technically and culturally complex, we planned ahead financially and administratively both inside and outside of the library. We made mistakes along the way. But the most important thing we did was make friends with time—working years ahead in three key areas.

Administrative Groundwork

Once 2020 was identified as an ideal migration window, the Library immediately began internal system migration conversations. In 2016, the Library first informally alerted the library staff, and then the institution’s senior administration, of the need to migrate systems. In 2017, the Library made a formal request for capital funding for the project, recommending a 2020 migration. This administrative groundwork helped the institution plan for major costs in certain fiscal years and to incorporate important pre-migration work, including the services of an external data consultant and new authority services. Inside the Library, it was the start of a three-year long conversation that we knew was as much about a culture as it was about a technical system.

System Data Review: The Gift of an External Eye

Very early in the process in 2017, the Library engaged an external data consultant, who was a systems and data expert from a nearby university. His expert knowledge of Voyager and his objectivity as an outsider was vital in helping to narrow our focus by asking good questions that challenged our status quo, unusual habits, and humorous picadilloes. He helped us view our data with fresh eyes and to identify what circulation, cataloging, acquisitions, and other policies, workflows, and matrices we should focus on in the following three years. The consultant also helped us clean-up and standardize some top-level data and delete unnecessary data. Later, when the time came to package the full Voyager database to send to Innovative for the Sierra go-live in 2020, he helped us with that delivery too.

Having an informed external eye on our data, several years in advance of migration, was absolutely critical to our process. We knew what to focus on and in what order. We had the time to review messy, historic data and to have complex (often very long) internal conversations about policies, procedures, and workflows. For example, the Circulation team completely reviewed and redefined all of its patron categories, rearticulating circulation and access policies (and related system matrices) along the way--ultimately moving from 69 historic patron categories to 15. By migration, we had a tight grasp on very clean data sets: we knew what our data types and categories meant (this was not true at the beginning of our process) and the data in each had been reviewed and cleaned. By 2019, when it came time for a functionality review of library systems in a series of dazzling sales demonstrations by the major vendors/platforms, we knew what we wanted from our new system relating to our major data types.

Physical Inventory

Parallel to its twenty-year residency on Voyager, the Library’s physical collection had not been comprehensively inventoried in over twenty years. Following earlier collection moves (partially off-site in Connecticut; other on-campus temporary locations) during the library building construction project in the 2010s, the large physical collection was due attention. The migration project provided an ideal time to begin a physical inventory project that, we knew from its earliest days, would extend beyond the new system go-live date. The collection was divided into sections and a detailed workflow was developed via a close collaboration between the Circulation and Cataloging departments. The inventory project passed its halfway mark in March 2020, when the pandemic required the Library building closure. Even with just half of the physical collection inventoried by go-live in July 2020, bibliographic data quality had massively improved: uncatalogued and miscatalogued items were corrected; some areas were modestly weeded; missing items were replaced or deleted; spine labels were corrected; some material that had been in the wrong place—either conceptually or shelved erroneously—was moved to its correct home. All of these corrections, of course, immediately aided our patron researchers as much as our migration team.

In the 2017-2020 migration, we learned to draw the piano keys on the table—sketching out a timeline and project milestones relating to administrative support and technical and cultural spheres of work. In our next migration, we’ll be ready to play those keys.

Playing the Piano Keys

As we prepare for our next migration, we’ll again capitalize on that most valuable resource, time, by planning at least three years out administratively and culturally, engaging an external data consultant early in the process for an objective lens on data priorities, and continuing the suddenly-paused physical inventory. We’re also doing several things differently.

1. Shift to Migration-Ready Mindset

The 2020 migration was very much a “project,” with a firm set of tasks and an end date. Leading up to and following the new system go-live, we decided that our mindset about system work and planning needed to shift: in the future, we need to remain migration-ready.

This conceptual shift has been helpful in communicating within the Library and with the senior institutional administration about the contemporary realities of more frequent system migrations for at least the next decade; the Library is unlikely to be on a system for twenty years again (and that is a good thing). The reasons for this are many and familiar: on one hand, the library system/platform marketplace is, and will continue to be, a volatile and in-development space as new technologies debut, corporate mergers continue to narrow the commercial field, and new types of vendors, partnerships, and collaborations evolve.

But even beyond those realities, other large-scale changes will impact what the Library needs and wants from its backbone system. Changes to the curriculum (locally or across theological education more broadly), to researcher strategies and practices, and to the future shape and skillset of the library professional workforce will have an equally important impact on system solutions. Going forward, we need to be ready to move systems quickly, and for different reasons than in the past. Remaining migration-ready is our new default.

2. Recommit to Data Quality

To remain migration-ready, we’ve recommitted to data quality and to bibliographic data excellence in particular. This commitment both extends the initial investment of time and resources from the 2017-2020 migration period into the future and positions us to move, connect, or integrate quickly with new systems or tools.

We’ve instituted an annual library system data review cycle, setting aside three weeks each summer for this work. Data custodians review data by type across the system, with an aim of keeping categories like item statuses reviewed and pruned. This work sometimes shines a light on a larger-scale issues that need more extensive attention, and those projects go on our bigger Data Priorities project list. Similarly, we’re restarting the physical inventory, albeit on a smaller scale and with a new design that we hope will result in a steady, iterative process across the collection, ongoing. It is a collection management ideal that we have never been able to realize.

And the Library has committed mightily to excellence in quality cataloging, knowing well how this work is at risk in other places. We’re keenly aware of how vital accurate, rich bibliographic metadata representing theological and religious traditions and scholarship will serve as an increasingly valuable contribution to researchers and other institutions as linked data and the semantic web grow.

3. Document Large-Scale Work that Must be Bequeathed to the Future

Our final change in approach was also a lesson learned from the 2020 migration. We’re identifying early and clearly what we won’t try to do and documenting it.

Preparing to move historical data off of a twenty-year system uncovered a few deep chasms of work that need more extensive time and attention, well beyond what we will be able to do in the next several years with current resources. We will only attempt to accomplish one of these large-scale projects prior to the next migration: the review and resolution of approximately 7,000 duplicate OCLC records in our system, a messy legacy of outsourced cataloging and uneven historic local practices that precludes automated handling.

For the other large-scale projects, we’re documenting them clearly in a single shared workspace. We’re describing the projects in lay terms, providing concrete examples, outlining the scope and extent of the work, articulating collection and research impact, and providing estimates about resources needed to appropriately tend to the projects. We’re building the list we did not have—a collection of the sometimes-hidden stories about the collection—for our future library and administrative counterparts.