3 ms·
Love the idea, but seems a little too fly-by-night for me to want to trust real production stuff to it. DDoS-ing DNS providers seems to be the rage these days,
by apendleton 14y ago
Love the idea, but seems a little too fly-by-night for me to want to trust real production stuff to it. DDoS-ing DNS providers seems to be the rage these days, and I'd hate to be collateral damage because the weekend project I was using as a DNS provider couldn't handle it...
And I totally get that this was just for fun... but a variant project that still used git as the UI, but sat on top of a bigger-player DNS provider's API (Route 53, maybe?) could probably give you the best of both worlds: a weekend's worth of hacking, but still using infrastructure people felt like they could trust.
- Amfy 14y agoThanks for your comment, and yes, of course I understand you. I would have the infrastructure (an anycasted network with 11 PoPs, etc) - but as you said, was only a weekend project and I wasn't about using this for the project. But hm, I could later just rewrite this away from a independend dns service to some parsing thingy to input zonefiles and use APIs of existing, reliable providers (most of the popular ones have nice APIs). It's something to think about, thanks.
- shykes 14y ago+1 on making this a frontend to 3d-party DNS providers. Git-push integration with Dyn, Route53, Zerigo/8x8, Gandi, etc. would be an awesome feature.
- mryan 14y agoI like the idea of building a bridge to other DNS providers - it seems like less work in the long run, for a much more reliable service. Using CloudFormation to manage your Route 53 records will achieve most of the goals of this solution, although editing CloudFormation stack templates has a little more overhead than managing zone files.
- wereHamster 14y agoI use git to manage my dns and publish to my server, which serves as a master for slave servers hosted by zerigo. Not sure if Route 53 can serve as slave servers.