4 ms·
This is all good stuff. AWS takes a while to grok but once you do, it offers so many new possibilities. The Aha! moment for me came when playing with SimianArm
by dmourati 13y ago
This is all good stuff. AWS takes a while to grok but once you do, it offers so many new possibilities.
The Aha! moment for me came when playing with SimianArmy, the wonderful Netflix OSS project and in particular, Chaos Monkey.
Rather than build redundancy into your system, build failure in and force failures early and often. This will surface architectural problems better than any whiteboard.
https://github.com/Netflix/SimianArmy https://github.com/Netflix/SimianArmy
Also, check out boto and aws cli.
http://aws.amazon.com/sdkforpython/ http://aws.amazon.com/sdkforpython/
http://aws.amazon.com/cli/ http://aws.amazon.com/cli/
- benjaminwootton 13y agoI think I saw a tweet along the lines of "don't think of EC2 instances as you would servers." The point is, achitecting for the cloud is all about treating your infrastructure as very transient and failure prone. Hopefully stuff doesn't fail but architecting for that gives you a much better application. Netflix are my favourite tech company based on their open source distributed system stuff. Check out Hystrix and their circuit breaker implementation. I am going to lean on that heavily in an upcoming project such that entire application tiers can dissapear with graceful degredation.