Somaya Langley is Digital Preservation Manager at the Science Museum Group. This is a report on two Birds of a Feather sessions at iPres 2026: Digital Preservation Storage Paradigms and Virus Scanning Workflows.
The Birds of a Feather (BoF) sessions – selected by the Ad-Hoc Committee Co-Chairs, Francesca Mackenzie and Sean Macmillan – were the highlight of this year’s iPres 2026 conference in Copenhagen. Encouraged by the DPC’s Dr Andy Jackson, I ran two sessions.
Chatham House rule
The BoFs provided the opportunity for digital preservation practitioners to discuss current challenges on specific topics, off the record. Both BoFs I facilitated were conducted under Chatham House Rule. No names of individuals or organisations are included.
Thanks go to Caylin Smith from Cambridge University Libraries and Marco Brunello from the Archaeology Data Service for taking anonymised notes.
Birds of a Feather: Digital Preservation Storage Paradigms
For over a decade, organisations have been moving their digital collections into cloud storage environments. Those with larger digital collections (i.e. PBs of data) are the exception. They have often opted to retain their existing Linear Tape Open (LTO) tape libraries and robots. Meanwhile, other organisations have chosen hybrid storage approaches, such as tape libraries coupled with cloud storage. After a number of years storing digital collections in cloud environments, selected organisations are even choosing to ‘decloud’. Discussions held as part of the Digital Preservation Storage Paradigms BoF is integral to our ability to Save our Stories (SOS); this year’s #WDPD2026 theme.
Background
While the 3-2-1 backup strategy has been fundamental to digital preservation and ICT backup for physical hardware infrastructure, how does this strategy map to the cloud storage context? Is 3-2-1-1-0 widely accepted? I first took this topic good practice digital preservation storage in cloud contexts to the 2023 DPC Members Unconference and Networking Event. No definitive answer emerged.
With an immediate need to draft requirements for preservation storage at the Science Museum Group – for both Working Space storage as well as part of a Digital Preservation System – I facilitated what turned out to be a popular topic, with 88 people in attendance.
Prompts were provided, and an engaging discussion emerged between the practitioners and vendors present. The ‘rule of three’ was still viewed as applicable, particularly in regard to storing data across different storage media (or vendors). The importance of air gapped storage to mitigate against cyber security threats was also emphasised.
Storage options, challenges and confidence in storage solutions
Storage options of LTO, on-prem, Network Attached Storage (NAS), public and private clouds, plus hybrid approaches were discussed. An audience member asked about what underlying storage technology sits behind the cloud. The assumption, from other audience members, is that cold storage is LTO. It was mentioned that vendors are often opaque about the storage technologies they employ. Another audience member raised the issue of some vendors not storing data where they say they are storing it.
The discussion then shifted towards how organisations could determine confidence that data was being stored according to digital preservation good practice. Questions were raised about whether the digital preservation community could assist each other in assessing and determining a ‘confidence level’ in where and how an organisation’s digital collections were stored.
One suggestion was for organisations (or a peer organisation) to carry out risk assessments to determine a confidence level. There was a mention of this being a potential activity for international digital preservation advocacy and membership bodies. Several audience members flagged that there already wasn’t available budget for membership fees at their organisations.
Cost concerns
Costs related to what might be viewed as ‘good practice’ digital preservation storage was also raised. There were concerns from those representing smaller institutions (with smaller budgets). If an agreed upon ‘good practice’ storage benchmark were set by the digital preservation community, they wouldn’t be able to meet the minimum requirements. Also mentioned were larger organisations who, despite their size, had small budgets for digital preservation. There were concerns about what this would mean if they were unable to procure the same storage hardware as their peers. A suggestion was that a ‘digital preservation community agreed-upon tiered approach’ was needed, based on both available funds and an organisation’s risk appetite.
Constraints and risks
Other concerns raised were the constraints imposed by organisations on their digital preservation staff and processes. For example, not being able to run workflows overnight or over the weekends. Related concerns were raised about storage infrastructure, including a question about what an organisation could do if they didn’t have the requisite tools or infrastructure to be able to write data to the available storage an organisation had. Digital Preservation practitioners were aware that each organisation needed to ensure that they had the right to a copy of the data if a cloud storage vendor ceased operating. However, a significant risk factor was raised which organisations typically don’t consider. If a vendor or service shut down, there would not be enough bandwidth for all the organisations to pull their data out. How we architect our infrastructure must consider this scenario.
Small-scale shared solutions and a cautionary tale
There were some good suggestions regarding the potential for small networks of institutions to hold a tape copy of each other’s digital collections. Alternatively, holding a second copy of an organisations data in an entirely different city was an example given as another solution. Cautionary tales were also reflected on, including the recent loss of data from Amazon Web Services (AWS) Bahrain (me-south-1) and the inability to restore data only held in these damaged Availability Zones, despite AWS’ ‘eleven nines’ durability. The importance of data replication and redundancy, balanced against data sovereignty, must be considered.
What we can agree on
Finally, we spoke about what we could agree on. The attendees agreed that one copy must be kept offline. This would be a separate copy to the digital content that would be accessible via a Digital Preservation System (or similar).
When I posed the penultimate question; “what good practice looks like”, the typical digital preservation answer was given. “It depends.” Audience members advised that their organisations would have different requirements – such as their internal policies and strategies – that could affect what would be possible. A further question was asked about whether there were different requirements for public cloud, local (private) cloud, organisation-managed hardware, or hybrid solutions. This is highly likely, but further in-depth discussion is needed.
Continued conversation
A one-off 45-minute discussion wasn’t long enough to glean definitive answers to good practice preservation storage. My takeaway is that the 3-2-1 approach in the era of cloud storage isn’t enough. It was also apparent that organisations shouldn’t be relying on only one commercial vendor for their storage. Additionally, organisations should have one copy of their digital collections stored offline.
It was clear that the conversation around good practice digital preservation storage needs to continue. Audience members said that they didn’t know what all their options were. One attendee suggested that an ‘idea catalogue’ should be created. Another suggestion was that it could take the form of a collaborative wiki, and that lightweight case studies should also be included. How we, as an international digital preservation community, can act on these suggestions is yet to be fleshed out. It’s obvious that we need a resource outlining the different storage options available, and documenting example case studies.
Birds of a Feather: Virus Scanning Workflows
The Science Museum Group has spent the past eighteen months developing and implementing a virus scanning workflow, however we need to learn how to handle the various scenarios we are likely to encounter.
Similar to the preservation storage BoF, the Virus Scanning Workflows BoF was well attended, with over 70 attendees in the room. Attendees again comprised of both digital preservation practitioners and vendors. Based on a set of prompt slides, a generous open discussion occurred. Several organisations had considerable experience and many examples to share. Many other attendees were keen to learn from others in the room.
Virus scanning around the world
Some organisations shared that they are carrying out virus scanning inconsistently, with other organisations advising that they are not doing it at all. Those who aren’t running virus scans were concerned it wasn’t taking place. They were hoping to learn from this session and then return to their institutions to implement necessary changes.
Organisations were often (but not always) running virus scanning on an isolated workstation, and/or as part of ingest into a Digital Preservation System. Quarantining (for a 30-day period) was mentioned – including how the timeframe was based on historical monthly virus signature updates to antivirus software – but it was not discussed.
One national organisation advised that Cybersecurity was being carried out at a national state, rather than institutional, level. Organisations who were routinely carrying out virus scanning within the digital preservation team advised that it wasn’t clear to them what to do once they had detected malware or a virus. A vendor mentioned that they were supporting large collections of emails, which included phishing emails, but were unsure of how these should be handled. Organisations also mentioned that they didn’t know how to handle virus scanning on large amounts of data (including video) and ‘big data’.
Virus scanning tool approaches
Two organisations mentioned that they (or their Cybersecurity teams) had suggested running three virus scanning software tools. One national organisation, who had been doing digital preservation for a long time, advised that different virus and malware software products could detect different malicious problems. An organisation advised that they had to stop running some virus scanning and malware services. The scans were not being performed locally, and instead were being sent (virtually) overseas, which didn’t meet their organisations’ requirements for data privacy and sovereignty.
Ingest workflows and workstations
More than one organisation mentioned that their ingest workstations were locked down. Another organisation mentioned that they had an older workstation for processing legacy media plus a modern ingest workstation. They found that their workstations were not meant for digital archiving, and there was a risk of ‘false negatives’ in their virus scanning processes.
One organisation recommended that several different ingest channels should be used, with some digital content requiring more attention than others. They mentioned that the ‘human factor’ was also a big consideration. An organisation posed that the legacy content contained fewer viruses, whereas another organisation advised that they were finding more viruses in the legacy than contemporary digital collections. More than one organisation advised that they were also currently using (or considering) transferring files via a file transfer service or SFTP. One attendee advised that they would not be using time-limited file transfer services (such as WeTransfer).
Risk profiles and false positives
Risk profiles were raised, however one organisation cautioned that some virus scanning software and systems determine risks from file format extensions (i.e. file extension mismatches), which digital preservation practitioners are well aware is not a suitable way for determining file types.
The discussion moved on to talk about how virus scanners behave, and specifically reporting ‘false positives’. It was agreed that a virus scanner’s default response to an ‘infected’ file was deletion. For digital collecting and preservation work, this is not a suitable reaction. One national organisation advised that if files were flagged as false positive, the only way to retain the file was to turn off the virus scanning in your tool chain, which isn’t a viable solution. Another organisation mentioned that their mandate was to safely store the digital collection, rather than to delete any files. They also advised that they had a high occurrence of false positives appearing in web archive files.
Working with (and around) your tools
Based on the experience of one organisation, a ‘sandbox’ workflow is required for retaining files that had been flagged as false positives, or files that contained viruses (and that an organisation determined as low risk). There was also a brief discussion about how to test your workflows to ensure virus scanning microservices were running in a Digital Preservation System, by using the EICAR file.
Collecting viruses
The discussion then moved on to collecting viruses. More than one organisation is actively collecting viruses. One European organisation wanted to know that their repository was able to securely keep the virus for the future. Another organisation collects spam emails to learn from them. While other organisations are keeping viruses to be able to research viruses and test antivirus solutions. An example was given of a USA-based organisation who chooses not to remove some types of viruses, such as certain types of macro viruses affecting legacy Microsoft files.
Virus scanning beyond ingest
Virus scanning was typically being run when digital content arrived at an organisation. Often, it was run again on ingest into a Digital Preservation System. Yet organisations hadn’t been considering rerunning virus scans before providing digital content to their end users. At least one vendor advised that this would be possible, however it would come at a cost.
One organisation advised that providing access to web archives now, were finding that linkrot meant more than just broken links. Links in websites, that were working when the website was archived, had ceased functioning over time. As a result, this organisation had found that expired domains had been bought up (often in bulk) and were now pointing to unsavoury locations. Rerunning virus scans, before digital content is provided to end users and/or at other points in the lifecycle of digital content (other than at ingest), has now emerged as a serious consideration. One attendee advised that both confidentiality and risk aversion were needed. Organisations must ask themselves if they are meeting security standards when rerunning virus scanning.
To be continued
The audience felt that they had learnt a lot participating in the discussion and that continued community awareness was needed. Some initial ideas of how to keep the conversation going were floated, now I need to follow up on that.
Wrap-up and going forward
I always come away from these discussions both awed by the amount of knowledge and experience in the room, as well as how many things digital preservationists must consider. It is affirming for myself and other participants to realise (and be relieved) they we not alone in our challenges. Both discussions felt like the starting point to a much needed and hopefully continued conversation. Let’s see what is possible in 2027.