I'm going to throw in my 2c here...
On Fri, May 24, 2013 at 9:33 PM, Morgan Blackthorne stormerider@gmail.com
wrote:
I don't know your requirements, but I do have to ask; is there a specific
reason you're looking at using OpsWorks instead of running Hosted Chef or
an Open Source Chef server? Until AWS upgrades OpsWorks to Chef 11, I think
it's a bit crippled, personally.
Even as someone who has mostly used Chef in a client/server environment, I
don't think it's fair to describe OpsWorks as crippled. It's old, sure,
but it's useful today - though I think Amazon are being incredibly
disingenuous in claiming that you can go ahead and use the community
cookbooks with it.
I looked at it for a project recently and decided that the older Chef
version would be an irritation, but not a deal-breaker.
Why not Hosted Chef? For one thing, it can be painfully slow in some
regions. My experience in ap-southeast-1 last year was not especially
positive. For another, it's not free. Though I think the pricing is
reasonable, cost sensitivities in bootstrapped projects make it a hard sell
(particularly when the quality of service is less that awesome).
Why not Open Source Chef? Not necessarily any cheaper than Hosted (if
you're running it on AWS), and it's something else to manage - that's a
small increment in many environments, but a big deal when you're standing
up a small environment.
Why would I consider OpsWorks instead of Chef Solo? There's a bunch of
plumbing involved in running chef-solo securely that I would have happily
avoided, and the "lifecycle events" in OpsWorks sounded like they'd be
useful enough to replace search for the parts I really cared about. No
(encrypted) data bags is a bit sad, but wouldn't have been an
insurmountable problem.
So, then, why didn't I use it? The dealbreaker for me was spotty support
for provisioning and managing other AWS services. I was also somewhat
uncomfortable with the level of control I appeared to be trading for
convenience, and wasn't confident that I'd be able to easily shift back in
the other direction. I ended up using CloudFormation with Chef Solo. It
has involved more plumbing than I'd like, but there's no deal-breakers (for
my use case) and I'm confident that I can easily upgrade to chef-client in
the future.
I can imagine using OpsWorks for other projects, but I'm not bothered by
the prospect of backporting community cookbooks (or implementing my own).
The cost of doing so isn't always significant, and the benefits are worth
something. That being said, folks who are fairly new to Chef might want to
wait until they AWS upgrade to version 11. The community largely moved on
from 0.9 a long time ago, and it's not a good place to start.
Zac