My Takeaways from the Library of Congress DSA Conference: Digital Preservation Is Entering a New Era

I was honored to be invited to present @Geyser Data’s perspective at the Designing Storage Architectures for Digital Collections conference at the Library of Congress in Washington, DC. The event brought together an exceptional mix of leaders from the Library of Congress, analysts, preservation practitioners, end users, and storage technology vendors for a serious discussion about the future of digital preservation.

What made the conference so valuable was the quality of the conversation. This was not a generic storage event. It was a working dialogue about scale, risk, econ

omics, sustainability, and the growing gap between what institutions need from long-term preservation and what traditional storage models are actually delivering.

A few themes stood out to me.

• First, digital preservation is no longer just a capacity problem. It is an architecture problem. Across the conference, it was clear that organizations are wrestling with how to retain control, maintain integrity, and preserve accessibility as collections grow into tens of petabytes and beyond. The challenge is not only where data sits, but how it is governed, validated, refreshed, migrated, and made usable over time.

  
  

• Second, AI is changing the meaning of archive. One of the most important ideas raised during the event was that data does not stay “cold” forever. As AI and machine learning become more central to institutional workflows, archived content increasingly becomes a source of future intelligence, enrichment, indexing, and reuse. That changes the design criteria for preservation. Archive can no longer be treated as a dead endpoint. It has to become a durable, trusted, economically sustainable layer that still supports future access and activation.

• Third, economic predictability matters as much as technical durability. A number of discussions highlighted the operational and budget challenges institutions face with large-scale cloud archival models, especially when retrieval, migration, and long-term access become uncertain or expensive. Preservation strategies cannot depend on economics that become punitive at the moment data is needed most. Long-term retention has to be designed for resilience not only in technology, but also in budget reality.

  
  

• Fourth, simplicity remains one of the most underrated requirements in storage. Modern tape continues to prove its value in durability, density, sustainability, and long-term economics. But one point came through clearly: when archival systems are too hard to operate, they create friction that limits adoption. The future belongs to architectures that preserve the economics and sustainability advantages of deep archive while removing the operational burden that has historically come with them.

  
  

• Fifth, sustainability is becoming measurable, not aspirational. The conversations around carbon accounting and storage efficiency were especially timely. Preservation at scale cannot ignore energy, water, and lifecycle impact. The right long-term storage architecture should reduce operational overhead and energy consumption while still improving trust, resilience, and access. Sustainability is no longer a side conversation. It is now part of infrastructure strategy.

I was also struck by how broad the innovation horizon has become. Alongside discussions on LTO, cloud preservation, and workflow automation, there was real momentum around DNA storage, optical approaches, silica-based concepts, and open long-term storage frameworks. Some of these technologies are still emerging, but they reflect an important reality: the market is actively searching for better answers to the long-duration data challenge.

My own takeaway is straightforward: the future of preservation will belong to architectures that combine durability, simplicity, control, and predictable economics.

That belief sits at the core of what we are building at Geyser Data.

Our view is that cold data is not dead storage. It is dormant strategic value. To unlock that value, long-term preservation platforms need high-performance ingestion, fixity at ingest, strong auditability, secure isolation, seamless migration and refresh, and the ability to support future AI-driven discovery and reuse. Just as important, they must do this without hidden retrieval penalties, infrastructure ownership burdens, or unnecessary operational complexity.

The Library of Congress DSA conference reinforced that the preservation community is asking the right questions. The next step is to make sure the industry answers them with architectures that are practical, sustainable, and built for the next decade, not the last one.

Thank you to the Library of Congress and the conference organizers for convening such an important discussion, and to the many speakers and participants who contributed insight, candor, and urgency to the conversation.