RightScale User Conference NYC 2011 -
Darryl Eaton - Director of Product Management, RightScale
Moving from traditional IT to the cloud involves a number of paradigm shifts at both the operational and architectural levels. How you build and manage your applications in the cloud requires a new way of thinking. For one, you don't fix a flaky server - you trash it while launching a replacement. And you don't maintain idle disaster recovery machines - you keep blueprints for those machines that you can spin up on the fly. By enabling you to "program" servers before they even exist, the cloud challenges you to think about IT service delivery in a novel way. We'll demonstrate how RightScale automation makes it easy to take advantage of the cloud's inherent agility.
5. Fixed Infrastructure Cost $ You just lost customers Predicted Demand Large Capital Expenditure Opportunity Cost Traditional Hardware Actual Demand Time
20. We hope to see you at our next RightScale User Conference! See all presentations and videos at RightScale.com/Conference.
Notas do Editor
Back in 1999, I was Director of Technology for a small startup called WiredPlanet. In that role, I was also the only system administrator for the company, running a deployment of servers, a NetApp box, and a couple Cisco switches. I had three different networks, all delineated by nicely color coded and custom-cut cabling. Everything worked great… for a while. Then one night, we switched over our office DSL line to a higher speed network, and our IP addressing scheme changed. Little did I know that someone had an NFS mount back to the old office network that was now failing, tying up the main database. To add to the drama – I had one of those famous dot.com boom parties to go to that night in the SF Moma, so this really struck at a bad time. I then proceeded to debug the situation. I had to ssh into a different machine and get into the database via the internal network since the external network was hosed. I then could not kill the NFS process, so I had to take a mysqldump and call Exodus to have them reboot the machine. After waiting for what seemed to be an eternity for the reboot, I logged in and tried to recover the database from what was on disk. It was corrupted, so I then had to restore from my backup – all manually. Finally, hours later, I had the whole system back up, and I was able to get to the end of the party… thoroughly fried. So, when I started at RightScale, I was enamored with the fact that I could have just hit “Relaunch” on that database, and everything would have been back up magically in 15 minutes. Or, to use a real example – I was on vacation when one of the Business Intelligence machines I had set up for RightScale went down. And someone else who had never seen it before logged into RightScale, pressed play, and all was well with the world once again.The reason this is possible is due to strategic paradigm shifts that the cloud offers that ultimately remove a lot of the infrastructure complexity we used to have to deal with. This is what we’re going to explore today.
Also operational efficiencies, locations of datacenters, etc.
Because they are blueprints:Spin up for testingSpin up for disaster recoverySpin up for releaseAlso need to realize: 100 servers x 1 hour = 100 hours x 1 server
While the cloud makes things more flexible, you’ll be surprised to find that many things are still fixed. Could computing management keeps these flexible too. Single cloud vs multi cloudOne learning ramp up; processes shared across cloudsExpensive Local Tools vs. Integrated SaaSSupport it yourself vs. professional services
Fix the the script in the ServerTemplate, not ad hoc on the server and then forgetting about it. Encapsulate and record best practices.
Take one of two paths to get started:Find what is most adaptable to the cloudFigure out what you hate most, and research if the cloud can solve it for you