Hacker News new | past | comments | ask | show | jobs | submit login
AWS EC2 Spot instances can now be stopped and started like On-Demand instances (amazon.com)
97 points by HedgedHuman 11 days ago | hide | past | web | favorite | 16 comments

Spot also supports hibernate on some instance types: https://aws.amazon.com/about-aws/whats-new/2017/11/amazon-ec...

Hibernate is just like good old hibernate from windows XP days. You save the RAM to disk. And resume your processes on next boot and they won't even notice they were stopped! (Except for the time has changed...)

It might be really helpful for low priority batch processing.

Disclaimer: I've never tested this.

Neat! Did not know that. Thanks :-) The 100GB RAM cap seems rather limiting though.

The ideal case scenario imho (one step beyond this stop/start-at-will feature) would be "convertible" spot - being able to migrate to on-demand with zero downtime when ec2 needs that capacity back - that'd solve a lot of my problems.

You can (effectively) do that already -- just set your spot bid to the maximum you're willing to pay, i.e. the on-demand price.

This isn't actually effective. When there are instance shortages and someone launches an on-demand instance, no matter how high you set your spot bid price your spot instance will still be terminated in favor of the on-demand instance. We run hundreds of spot instances with their max price set higher than on-demand and still see terminations on popular instance types: there are few day-to-day, but periodically nearly all of them are terminated at once. During those situations, if you went to replace the terminated spot instance with an on-demand instance, you'll even get an error stating that they're out of on-demand capacity for the instance type in that zone, so it's pretty clear that it happens when they've completely run out of capacity for some instance type in a zone

Spot on!

Some AZs don't have much capacity for certain types of instances.

Fun fact: An availability zone A for person 1 is not guaranteed to be the same as availability zone A of person 2. AWS mixes them up per account.

When a spot instance gets interrupted, they've supported terminate, stop, and hibernate for a while. Hibernate is effectively just stop, but has an additional hook inside the instance to tell linux to save memory to disk while shutting down.

This post is different: with on-demand instances you can stop them whenever you like and only be charged for storage while they're in the stopped state and then start them up when you need them again (keeping all the same settings, like the IP, etc). A stop/start is also the way to effectively move your instance to new hardware if there is an instance with the underlying host. Annoyingly, the stop/start API calls were rejected for spot instances. This announcement means this is now finally possible.

Very cool, been running into issues with this previously trying to manage mixed instances kinds in an asg.

One tip, as an easy way to track api changes in aws, checkout https://awsapichanges.info/

Aws spot instances have saved me so much money... it’s such a great service.

Rapidly moving toward compute as a commodity priced in the same way global commodity markets are. Interesting times

When can I start trading options on EC2 pricing though??

If Amazon or wall st. can figure how to make a cut off the trading. At that point they will trade options. Sounds like something I would want to trade... kinda that whole Enron thing... (except hopefully the fraudulent part)

When EC2 becomes an actual market, rather than a monopoly?

This is a useful addition.

It's for persistent spot requests, so presumably it doesn't include defined-duration spot instances. Perhaps that will come later.

Perfectly possible to work around it even if not, of course.

Obviously a response to yesterdays post:


Huh? They’re completely unrelated.

No, they're two different things.

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact