This would help contribute to 0 downtime for me if possible. As far as I know it is not possible, but wanted to be sure.
-
No. The closest thing is probably Auto-Scaling aws.amazon.com/autoscaling– LinuxDevOpsApr 26, 2014 at 19:24
-
2@TheInternet your question was likely downvoted because it showed very little input on your part. Have a read through the following meta.serverfault.com/questions/3608/…– Drew KhouryApr 27, 2014 at 4:05
-
Thanks for censoring me @Drew Khoury. This type of question isn't a "what did you try?" type of question but a "yes/no" type. Take a look at "How can I get people to fully read my question" meta.serverfault.com/questions/2177/…– The InternetApr 27, 2014 at 23:44
-
It sounds like your real problem/question is around achieving zero (or close to) down-time which you can certainly do with AWS and EC2s.– Drew KhouryMay 10, 2014 at 3:49
2 Answers
You'll need to stop your EC2 and change the instance type to add more RAM so no it can't be done dynamically, or on-the-fly.
Alternatives
You're interested in no downtime so you should take a look into a number of features:
- Elastic Load Balancing (ELB)
- Elastic Network Interface (ENI)
You can create an ELB and attach an EC2 to it. Next you could clone your EC2, change it's instance type, and bring the clone into the load balancer. Finally you terminate the old EC2 out, completing the 'zero downtime' upgrade.
You could attach an ENI to your EC2. Create a clone, change instance type, attach ENI to the new EC2. Finally, terminate the old EC2.
-
3It's interesting that Virtualbox supports Hot Add Ram and Hot Plug CPU. Why not a cloud provider? May 29, 2014 at 9:04
-
@CMCDragonkai Consider that the instance might be running on a physical hypervisor that has no available RAM or additional CPUs. They could live-migrate the VM to another physical host, but I don't believe that AWS does this at all (or "instance retirement" for EC2s without ephemeral storage would not be a thing). I don't think it makes sense for AWS to offer hot-adding resources if they can't guarantee that it will always be possible.– cdhowieSep 28, 2019 at 16:01
-
Stopping, 'adding more ram' by changing instance type and starting again introduced me to a hell of errors and troubleshooting, couldnt ssh anymore, status checks failed for no clear reason etc. This seems like a bug and not much info about it. I ended up starting from scratch– WestJul 7, 2022 at 14:17