Does anyone else here actually support the whole concept of pay-per-hour? I thought people here would be all for it but it got a cold response.

Thanks for asking. Yes, a while ago significant progress was made into metered billing, however it was archived due to a change in priorities at the time and was never revisited. We've been discussing reviving that effort recently and hearing interest like yours helps raise its priority. Stay tuned.

Thanks,
-Chris

If it's for scaling up and down with load as Guspaz suggested, I can support it. Otherwise, I personally see no use for it.

_________________
Kris the Piki Geeker

I'd like it, the combination of stackscripts, the api and hourly billing would make some tasks I have to do periodically cheaper. A few times a month I have to spawn up a bunch of hosts for a few hours, currently I do this elsewhere.

_________________
Paid support
How to ask for help
1. Give details of your problem
2. Post any errors
3. Post relevant logs.
4. Don't hide details i.e. your domain, it just makes things harder
5. Be polite or you'll be eaten by a grue

As long as Linode doesn't become metered only then it wouldn't affect me.

If the price per day ends up being the same, why would being metered only be a problem? If you wanted to just keep a linode running 24/7, the bill at the end of the month would be the same. But if you needed to spin up a linode for whatever reason for just a few hours, you could do that and save some scratch.

Then again, I'm assuming that the primary difference between the current approach and "metered billing" is just to change the billing granularity to hourly instead of daily. If it also means a linode comes with no bandwidth and you pay for that as you go, well, I'd have to think some more about that. I suspect per-gig costs would end up higher, since the current per-gig costs make the assumption that most people don't use the full allocation.

I wasn't suggesting Linode abandon their current billing system because it works great for normal machines. I was suggesting per-hour billing as an optional extra way to buy short lived Linodes. Predictable costs are a big win for Linode and a big loss for Amazon.

Say the cheapest Linode costs $20 a month, the hourly cost for the same machine could be $0.04 an hour so it's at a 40% or so premium to cover the extra support, server load, and admin costs.

I was only suggesting using existing capacity that's currently idle for these short term allocations. Last time I checked Linode had a serious number of Linodes sitting idle that could be used for something.

Why do you suggest punishing people who are willing to use the infrastructure with a premium percentage? Reminds me of EC2, absolutely sucks as you have to pick the right hardware, the right reserved instance, etc.

Digital Ocean does it at the same price as the monthly price. It's simple. I know I'm getting the same cheap price regardless of when I decide to spin up an instance, or when I decide to kill it.

Personally I'd like to see Linode do a billing system similar to Google Compute Engine - minimum 10 minutes and pay by the minute. Keep it simple, same pricing structure as the monthly plans.

I'm late to this party, but yes - I'd love this. Actually I'd like per-minute better. And added to it, I'd _love_ the ability to store images somewhere (for a fee of course). If we had all of this, we could actually run a staging environment on Linode, but only run it for, say, a couple of hours a day as we needed it.

About a week ago I wanted to play with LDAP replication between different countries. I just can't do that at home unless I add some kind of network delay setup and buy faster disks.

I kicked off 3 small machines on digital ocean, set it all up, tested for about 30 minutes, and tore it all down. It was convenient and only cost 2 US cents. It would have cost about the same on EC2.

Per-hour machines are great for testing.