7 ms·
Yep, the amount of incompetence in Parler's case is just mind-boggling. Their failure to make it cloud agnostic being 1 (even though they claimed before they w
by nacs 6y ago
Yep, the amount of incompetence in Parler's case is just mind-boggling.
Their failure to make it cloud agnostic being 1 (even though they claimed before they went down that their app could run without AWS without issue).
Their failure to be incapable (still) to get it running again on non-cloud hardware..
And of course their inability to, as you suggest, move the frontend routing to a CDN/Cloudflare/DDosguard type service and have all the heavy lifting continue to be done via AWS/Azure/Google till you can run 100% on your own hardware.
- paul_f 6y agoMaybe they didn't think they needed to. Before a few months ago, you didn't yank someone off your network because you didn't like their politics. (yea yea, terms of service something or other)
- 4cao 6y ago> Before a few months ago, you didn't yank someone off your network because you didn't like their politics. AWS dropped WikiLeaks as a customer back in 2010 for political reasons, and apparently this was announced by a US senator even before AWS did so themselves: http://edition.cnn.com/2010/US/12/01/wikileaks.amazon/index.html http://edition.cnn.com/2010/US/12/01/wikileaks.amazon/index....
- zxcvbn4038 6y agoCloudflare and Daily Stormer is another good example. The mistake there was the implication that Cloudflare was endorsing the stormer rather then being a neutral carrier and Cloudflare shut down that idea quickly. I have to applaud them, it’s hard to be neutral in the US, everything is “with us or against us.” People watch too many gangster movies I think.
- SirSavary 6y agoNot that my company will produce anything as controversial as Parler but I now use this example to explain why we go the extra mile with things (e.g. scrubbing headers). Sometimes the worst-case scenario really does come true.
- rualca 6y ago> Yep, the amount of incompetence in Parler's case is just mind-boggling. Parler's case has zero to do with incompetence, and it boggles the mind of how suddenly putting together web apps with bare metal servers is somehow depicted as normal or usual or basic competency skills. It isn't. At all. The majority of the world's web apps are hosted by one of the four major cloud providers. AWS alone hosts about 1/3 of the world's websites. The only reason anyone ever had to move away from AWS ever was due to their high prices, and even then the decision was always to move to another major player. Hell, whenever I point out that there are bare ones cloud service providers which are far cheaper.than AWS, such as Hetzner, I'm faced with at best an industrial dose of skepticism. Let's face it: up until now there wasn't any reason at all to even consider hosting your startup app on a non-major cloud provider. At all. In fact, it's tremendously hard to argue using anything beyond them. You need scale on-demand, high-availability, and reliability. You do not get any of that by even picking most of the non-major cloud providers, let alone running bare metal servers. Post-facto accusations of incompetence are not justifiable, specially when criticising the undisputed go-to rule that driver over half of the world's websites.
- paulie_a 6y ago> bare metal servers is somehow depicted as normal or usual or basic competency skills. It isn't. At all. Just calling them servers is sufficient. I don't know when this "bare metal" crap started Standing up a service is a basic competitancy