Hacker News new | ask | show | jobs
by zdragnar 2918 days ago
Throwing it in a git repository ensures that every person who clones it can readily republish it. Hosting said repository on Github, when anticipating an utterly massive spike in traffic, is an easy way to not have to pay for said spike in traffic (either from provisioning or data transfer).

I didn't look at what format the "database" is in, or if the size would make it (im)practical to simply zip it up and email it around, but if the format isn't readily consumable by non-technical people, there wouldn't be any reason to not utilize a tool like git anyway.
1 comments
JetSpiegel 2918 days ago
Why not host the repo on his site? At least a read only copy.

`git clone --bare` is enough.

link
zdragnar 2917 days ago
Git itself was a bit of a red herring, even though it was somewhat relevant to that specific point. I know nothing about what hosting platform the author is currently using, so to make a quick assumption:

- hosting on AWS is not free for super high traffic (assuming the free tier can't keep up) - serving files from S3 is not free (though it's cheap enough at low read levels, it adds up)

At a typical level of traffic, the author's current host may be sufficiently inexpensive. Assuming the author was assuming many, many times the usual traffic (even if everyone is kind enough to bare clone), it would be a pointless expense.

Of course, third party hosting can take the content down... and this is where git became relevant. Assuming the author was more interested in distributing the content than the prestige of being the distributor, even though Github etc. took down the repo, every person who has since cloned is now capable of re-publishing to any new upstream repository of their choosing, on any server.

Assuming, again, that all of this was the goal, it probably made sense to utilize the free, fast, scalable third party hosting as long as possible rather than risk self-hosting slowing down or collapsing under traffic, or creating a massive spike in cost.

That's a whole boat load of assumptions, any of which could be wrong. In the realm of possible motivations, though, I think it's a fairly logical conclusion.

link