DigiFilm develops a new approach for long term data preservation

DigiFilm Corporation, a french entity started in 2013 with the mission to develop a long term data preservation media, participated yesterday to the 59th IT Press Tour organized in Valletta, Malta. The team is pretty small but very active and highly skilled with Rip Hampton O'Neil who leads the technical part, Antoine Simkine around innovation and Pierre Ollivier, CEO.

The session was presented by Antoine Simkine who introduced their technology named Archiflix. It is a WORM media in a "Store and Forget" philosophy. First it's about the media and the choice relies on movie film reel who has a very high longevity and robustness. Second, the reader must be a standard with classic commercial scanner. Third, unlike tape and some other media, the goal is to avoid any technology refresh or migration. Fourth, the sustainability constraint is a must over a multiple decade lifecycle.

Some tentatives have been used to store information for multiple decades but the restore or retrieve of information results in a transformation and a new generation. It means that the final data is not equal to the original one and also lost all associated metadata. Such development must deliver the capability to retrieve the exact copy of the data and all metadata.

The team invents the Pixa code format to encode, store and decode the source data and promote it as an open standard. When the access is needed, a simple scan plus a decoding phase are required and thus the perfect copy of the original file. Each reel embeds information about the algorithm to use for decoding, scanner characteristics and blueprint as well. So the reel is self or auto described and allow easy data access. The addition element resides in the Archiflix Vault, a specific location to keep all reels.

The company is in the process to raise a small round to continue the development of Archiflix and is looking for partners to accelerate this phase. The idea is not to build any hardware scanner or machine. We understand that the pricing is based on Pixa. In terms of performance, don't expect to be fast and it is not an issue as we speak about archiving, it shows 5000-10000 documents for 5 minutes, 200GB per 600m/reel. This kind of support will need 10 minutes for one reel.

DigiFilm plays against Piql, another company from Norway who market microfilm for a few years to store data. We also met Cerabyte, BioMemory and other DNA storage actors and even Folio Photonics, not ready yet. And we also know that some hyperscalers have started some initiatives. All this illustrates a real need. We'll see how this project evolves but for sure the storage industry needs sone iteration on these directions.