It's about https://www.cip-project.org/ which aims to support "civilian infrastructure" deployments (e.g. power plants) with a lifetime of 10-100 years (various upper and lower bounds are given, those seem to be the min and max).
The plan for accomplishing this is with a "Super Long Term Support" kernel and small core set of packages, and collaboration with Debian and some other communities such as OpenEmbedded (about which there was another DebConf talk). Or this is the start, because maintaining even a SLTS kernel for over 10-20 years doesn't seem feasible.
This seems like an excellent initiative, and I'm glad that they chose to work with Debian.
The talk didn't cover how computer systems in civilian infrastructure have been maintained over the past 50+ years. I'd love to know about that, and what CIP is taking from that history, if anything.
The talk didn't cover how computer systems in civilian infrastructure have been maintained over the past 50+ years. ...
I asked myself the same question. My (uneducated) guess is that up to now these infrastructure-type computer systems have been maintained in house; this may be increasingly unsustainable. The CIP project aims to pool resources so that common components can be jointly maintained for a longer term and at a less cost. A very admirable initiative!
Have you heard about the ancient IBM mainframe at my employer that replaced the Sperry Univac? We run DB/2 now barely. The actual databases date back to the late 1960s and are still in operation.
The employer has been thinking about porting the whole mess to Java and even got a couple patents for porting techniques. The efficacy of a US Government bureau patenting its own work made on official time is a discussion for another day. We keep being told we'll retire the mainframe "some day". Congress keeps giving us money to retire it but it never seems to be enough.
There is a reason it is called The Martinsburg Monster. Why the current-day backups still sit in Memphis and Detroit seem to be things that elude me as to organizational thinking. You'd think we would plan better in situating CIKR.
Stephen Michael Kellat at 2018-09-11T03:34:53Z
Tyng-Ruey Chuang likes this.