Adrian Cockcroft, Director, Cloud Architecture at Netflix posted the following presentation, Performance Architecture for Cloud on Slideshare yesterday. In this presentation, he shares the concerns Netflix has with AWS Load Balancer limitations, and how SimpleDB needs a memcached front end in addition to several other limitations Netflix has found. He makes additional recommendations through the presentation deck that are worth checking out given the extent of the AWS outage over the last four days. His brief but insightful coverage of Performance Tools Architecture and assertion that AWS is challenged by data center tools and metrics is prescient given the outage that occurred.
Of the hundreds of blog posts, videos and stories filed by industry and mainstream media on the AWS outage, the following are the three best I’ve seen. Ray Wang’s post, Monday’s Musings: Lessons Learned From Amazon’s Cloud Outage is excellent along with George Reese’s thorough and excellent blog post AWS Outage: The Cloud’s Shining Moment and Phil Wainewright’s Seven lessons to learn from Amazon’s outage. All three posts are excellent for getting informed perspectives on the AWS outage and how companies need to plan and respond.