Preservation and Access to Databases: Why There’s No Silver Bullet 

Databases sit at the heart of modern organisations. From clinical trial data and manufacturing records to training systems and environmental monitoring, they underpin critical operations and regulatory compliance. 

But when it comes to long-term preservation and access, databases present a unique challenge. 

In an Arkivum webinar, Matthew Addis – CTO and founder explored why preserving databases is so complex and what a practical, risk-based approach looks like. 

What Do We Mean by “Database”? 

A database is so much more than a collection of data. It combines structured information with the application used to manage and access it. 

In most cases, this involves three interconnected layers: the application users interact with, the database structure itself (tables, schemas, queries), and the underlying storage. Each layer introduces its own preservation challenges, which is why simply keeping the data is rarely enough. 

The Core Challenge: Data + Functionality 

One of the biggest misconceptions is that database preservation is just about archiving data. In reality, preserving the ability to use that data is just as important. 

Databases are inherently dynamic. Users need to search, filter, and interpret relationships within the data. Static formats like PDFs cannot capture this. At the same time, the systems that enable this functionality don’t last forever. Software becomes obsolete, licences expire, and the people who understand the systems move on while retention requirements often stretch for decades. 

Regulatory Expectations: A Risk-Based Reality 

Regulated environments make this even more complex. GxP guidelines require data to remain complete, readable, and accessible over long periods of time, often 25 years or more. They also emphasise maintaining data integrity and, in some cases, preserving the dynamic nature of data. 

Importantly, regulators recognise that preservation is not a one-time activity. Over time, technologies change and perfect preservation may not be possible. Instead, organisations are expected to take a proportionate, risk-based approach that evolves throughout the data lifecycle. 

There Is No Single Solution 

One of the key takeaways from the webinar is that there is no single solution to database preservation. 

Some organisations take a “museum” approach, keeping old hardware and systems so they can be restarted when needed. Others migrate data between systems over time, while some capture entire environments as virtual machines. Another option is to extract the database and store it in open, standardised formats, reducing reliance on original systems. 

Each approach has strengths and weaknesses. For example, preserving original systems maintains full functionality but is difficult to sustain. Migration keeps data accessible but introduces cost and risk over time. Extraction into standard formats improves sustainability but may not retain every feature of the original system. 

In practice, organisations rarely rely on just one method. 

Why Format Choice Matters 

The choice of format plays a critical role in long-term preservation. Proprietary formats can introduce risks such as vendor lock-in or lack of future support, while widely adopted open formats tend to offer greater longevity. 

However, even formats considered “low risk” today may not remain so in the future. This is why ongoing risk assessment and active management are essential parts of any preservation strategy. 

Practical Approaches to Relational Databases 

There are several practical ways to preserve relational databases, each with different trade-offs. 

Database dumps can capture the structure and content of a database in SQL form, making them useful for restoring systems, but they are not always easily portable between platforms. Exporting tables as CSV files provides a simple and accessible way to view data, though it loses relationships and richer functionality. 

Migrating databases into open systems such as SQLite offers a more balanced approach, retaining structure and enabling querying in a widely supported format. Going further, formats like SIARD (Software-Independent Archiving of Relational Databases) are specifically designed for long-term preservation, capturing not just data but also structure, metadata, and provenance in a vendor-neutral way. 

A Combined Approach Works Best 

In reality, the most effective strategy is usually a combination of approaches. For example, an organisation might use a preservation format like SIARD as a long-term master, maintain an accessible version in SQLite for querying, and keep simpler exports for quick reference. 

This layered approach allows organisations to balance accessibility, sustainability, and risk, while ensuring that different user needs are met over time. 

Plan for Preservation Early 

Many preservation challenges can be avoided by planning ahead. Too often, organisations discover too late that their data is locked into systems that cannot be easily exported or preserved. 

By considering long-term retention requirements during system selection, organisations can ensure that data can be exported in open formats and validated as part of normal processes. Building preservation into data management planning from the outset makes a significant difference. 

Final Thoughts 

Database preservation is not easy, but it is achievable. It requires recognising that databases are more than just data, and that preserving their usability is essential. 

There is no single solution, but there are proven techniques. By taking a risk-based approach and combining multiple strategies, organisations can ensure their data remains accessible, trustworthy, and useful for years to come. 

Digital preservation is ultimately about maintaining value over time and with the right approach, databases can continue to deliver that value long after the systems that created them are gone.