Hi Robert, thanks for opening this discussion. I am one of these researchers (or data hoarders? :)) that looks a lot at historical data, so I appreciate that there are no plans to delete data (yet). Moving old data to cheaper/slower storage is a good idea imho if it reduces the cost for the RIPE NCC. It would be nice, however, to retain the ability of fetching results (or metadata) for specific measurements or timeframes via an API, although this would still require an active index. For me, it would be fine if this access is slow. I am saying this since the alternative is usually to put archive dump files on some FTP server, but I can not think of a suitable structure, e.g., for Atlas measurements. One file per measurement does not work for long-running, ongoing measurements. One file per measurement containing results for some timeframe probably blows up the number of files. Anyways, I support this proposal, but would prefer to retain a systematic (if slow) access to the data. Best, Malte P.S.: It may be obvious, but I mostly work with Atlas, less with RIS, so my comment is mostly related to Atlas data.