6 ms·

There's always a danger when you rely on outside provider for something that's critical to your business or life. Sometimes you can't avoid it or it's complete

There's always a danger when you rely on outside provider for something that's critical to your business or life. Sometimes you can't avoid it or it's completely infeasible to, but choose carefully. I see too many companies using AWS for stuff they could fairly easily self-host.

ceejayoz 9y ago
By "self-host", don't you mean "rely on a landlord / ISP / colocation facility"?
- gorkonsine 9y ago
  No, if it's for non-web-hosting stuff, I mean buying your own servers and administering them yourself. Obviously, web-hosting stuff is a little different. AWS is a lot more than just web hosting: it's a cloud computing provider.
- e59d134d 9y ago
  Back in days we rented servers and created our env, it was not really that hard. And it was super easy to move from one provider to another.
  - ceejayoz 9y ago
    > Back in days we rented servers As you do with AWS... > and created our env As you do with AWS... > And it was super easy to move from one provider to another. Which is easier on cloud hosting than it ever was when dealing with colocation providers.
    - raesene6 9y ago
      Well it's not quite the same with AWS if you start making use of any of their facilities past straight EC2 instances. AWS provides a number of SAAS offerings which, if you make use of them, tie you to Amazon specifically, as opposed to an IAAS offering which is likely to be more easily portable to another provider.
  - lotyrin 9y ago
    AWS created a Cambrian explosion of people hosting stuff that have no business doing so. Those people can't just go back to colocating hardware at datacenters... I wonder if they could even effectively get their equipment racked and wired up. Having to get approval for a capital expense was a filter that prevented teams with zero ops experience from trying to just brute-force google their way through. The teams I see the most often, in pre-cloud days got nothing done because touching a server meant going through experienced ops teams that could conduct architecture reviews, security reviews and maintenance planning but which didn't have any motivation or incentives to actually launch new products. Cloud means these teams can get things done, at the expense of not having any thought put to architecture, security, or maintenance. Sure, it's not hard, but it's not possible for them.

ceejayoz 9y ago

By "self-host", don't you mean "rely on a landlord / ISP / colocation facility"?

gorkonsine 9y ago
No, if it's for non-web-hosting stuff, I mean buying your own servers and administering them yourself. Obviously, web-hosting stuff is a little different. AWS is a lot more than just web hosting: it's a cloud computing provider.
e59d134d 9y ago
Back in days we rented servers and created our env, it was not really that hard. And it was super easy to move from one provider to another.
- ceejayoz 9y ago
  > Back in days we rented servers As you do with AWS... > and created our env As you do with AWS... > And it was super easy to move from one provider to another. Which is easier on cloud hosting than it ever was when dealing with colocation providers.
  - raesene6 9y ago
    Well it's not quite the same with AWS if you start making use of any of their facilities past straight EC2 instances. AWS provides a number of SAAS offerings which, if you make use of them, tie you to Amazon specifically, as opposed to an IAAS offering which is likely to be more easily portable to another provider.
- lotyrin 9y ago
  AWS created a Cambrian explosion of people hosting stuff that have no business doing so. Those people can't just go back to colocating hardware at datacenters... I wonder if they could even effectively get their equipment racked and wired up. Having to get approval for a capital expense was a filter that prevented teams with zero ops experience from trying to just brute-force google their way through. The teams I see the most often, in pre-cloud days got nothing done because touching a server meant going through experienced ops teams that could conduct architecture reviews, security reviews and maintenance planning but which didn't have any motivation or incentives to actually launch new products. Cloud means these teams can get things done, at the expense of not having any thought put to architecture, security, or maintenance. Sure, it's not hard, but it's not possible for them.