Here are some short stories I want to share.
- Common misstakes in cloud is over provisioning: server idling, inefficient utilisation monitoring. In short, you pay for resource that you don't use. Missing a data cycle handling.
- Everything can and will fail. Design your service according to that. Have a plan B for each of them. I had recently a dramatic Elastic Load Balancer failure that only caused 50% traffic fails. Mitigate it by change DNS and by pass ELB.
- To detect above issue, you must have something outside Amazon to monitor it.
- Amazon has 88k employees and army of engineers. Maybe they are not best on everything, but don't think you are smarter and better.
- If NasdaqOMX can adapt cloud for financial services, so can you. Security is not an issue. It is rather how you design and handle it.
- Maybe at 10% of time I miss the data center and hardware work, but 90% of time, I don't.
- Private cloud doesn't really scale down. CAPEX is still there. There are some benefits, but it is not enough imo.
- RTO & RPO and DR design is much easier when you know everything can and will fail.
We have not yet seem the full power of cloud yet.
No comments:
Post a Comment