19 Jan
2020
19 Jan
'20
3:04 a.m.
On Jan 18, 2020, at 1:53 PM, Steve Nickolas
<usotsuki(a)buric.co> wrote:
On Sat, 18 Jan 2020, Adam Thornton wrote:
A possibly-stopgap-but-maybe-not-idiotic solution just occurred to me.
This isn’t exactly the sort of data that’s GOOD for hosting via Git, since it’s
mostly-binary and mostly-read-only.
But checking it into a GitHub public repo until we were told to stop…or checking in some
indexes, paying for some S3 buckets, and using Git Large File Support pointing at the
buckets for the binary blobs…might not be the dumbest idea in the world. I mean, GitHub
is well-indexed, has a lot of incentive to maintain good backups, and probably isn’t going
anywhere any time soon.
Adam
So…that’s not a lot of archive, so I’m guessing
that it’s outbound bandwidth that will be the driving cost. But even that…how popular is
it _really_ ?
It seems like, given the nature of the collection, it might not be hard to persuade one
of the cloud providers into discounted rates for hosting, although…it’s so small that that
might not work, because that little data, well, you’re not a customer big enough to have a
Google or Amazon rep.
I’ll put out some feelers. Rough bandwidth data, if we can figure out some way to find
it, would be good to have.
What about renting an OVH server and slapping it on that? I think the KimSufi 1 is 500
GB disk space - dunno if that's cheap enough.