5 ms·

One thing I hate about the cloud providers is that there isn't an option to set a maximum cost. I would prefer to plug the cable of my side project than just re

by kuu 5y ago

One thing I hate about the cloud providers is that there isn't an option to set a maximum cost. I would prefer to plug the cable of my side project than just receive an email saying me that next bill is going to be over my cost. I understand not everyone would like to do that, but I would like to have that option.

herodoturtle 5y ago
I assume you meant “pull” or “unplug” the cable :)
- kuu 5y ago
  Yes ;)
cma 5y ago
They'd rather refund small guys for mistakes than give big guys an easy limit to set.
- kuu 5y ago
  I guess big guys don't want they service to suddenly stop, so they probably would not use this... But it's just a guess
  - electroly 5y ago
    Absolutely that. Storage costs money, so in order to absolutely cap your spending they would have to delete all your stored data, too. Deleting S3 buckets and EBS volumes on a spending blip is absolutely the last thing any company with any budget at all wants to happen, ever. It would be preferable for that not to even be possible in any situation. This is the sort of thing that only extremely small casual users want, and it isn't worth it to AWS to cater to those users. For everyone else, more complexity than a "kill everything at $X" switch is needed, and that's exactly what we do have. We don't get to absolutely cap our spending to the penny but we also don't risk having our data vanish because of a billing issue.
    - ghaff 5y ago
      I assume the sensible implementation would be cut off access and give you some period to settle your bill before the data is deleted.
    - CodesInChaos 5y ago
      > Storage costs money The dumb solution for that is to exclude persistent storage from the limit. The nice solution for that is supporting both "runrate" and "consumption" limits. Using a runrate limit, spinning up an instance, creating a file, etc. allocates budgets for running it continuously, which is released when shutting it down/deleting it. Hitting the limit prevents new resources from being allocated, but keeps existing ones alive. This should be used for persistent storage and instances used to handle base load. Using a consumption limit, the resource is shut down when the limit is hit. If the shut-off is delayed, the cloud service eats the overage, since they control the delay. This should be used for bandwidth, paid api-calls, and auto-scaling instances. The user should be able to create multiple limits of each kind, and assign different services to such limits. Alerts when going near the limit can help the user raise it, if that's their intention. For consumption, it might also make sense to have rate limiters, which throttle after a burst budget is exceeded, similar to how compute works on T instances on AWS. But those probably only make sense for individual services, not globally (e.g. throttle an instance to 100 Mbit/s after it exhausted its 5 TB/day bandwidth allocation, or throttle an API to x calls/s).
  - cma 5y ago
    For background batch jobs and analytics etc. they might want caps. Say something like a video transcoding workload. And lots of things could benefit not from a cap, but some kind of gradual degradation in bandwidth/instance allocation + a warning so you can raise the limits, it doesn't have to just shut everything down immediately using a hard cap.
frameset 5y ago
But there is an option. In Azure you can "set a budget". He even goes over it in the post. Did you read the linked article?
- lkxijlewlf 5y ago
  https://news.ycombinator.com/newsguidelines.html https://news.ycombinator.com/newsguidelines.html
defaultname 5y ago
Oracle has fantastic budget tools. Not just "you've passed your budget", but "you're forecast to pass your budget in 22 days before the month is up". And you can couple it with quotas to create hard budgets. AWS has decent tools in this regard, but it pales compared to Oracle. Azure is a product I've never used with any scale (just small projects), but the fact that it actually costs money to setup alerts is gross (and morally reprehensible). Even if it's a trivial amount, that alone just sours the product in my eyes. I mean, already Azure is pretty uncompetitive unless you're running on free credits, as Troy apparently is (purportedly some $13K per year, so unsure what the pitch for donations to cover a bill is about).
- schemescape 5y ago
  This piqued my interest, but a few quick searches (using a search engine--the Oracle Cloud site search only turned up press releases...), indicate that quotas just prevent you from spinning up new instances. That's helpful, but I was hoping for some sort of way to cap my bill (for hobby projects), even if that requries deleting resources. Oracle Cloud has an enticing free tier, but I'm too afraid to use it because it requires a credit card and I don't see any way to put a monthly cap on my budget. (I'm sure hobby projects with ~$5 - 10/month budgets isn't their target market, but I can dream :) Edit to add the page I was reading: https://docs.oracle.com/en/cloud/get-started/subscriptions-cloud/mmocs/quota-breach-details.html https://docs.oracle.com/en/cloud/get-started/subscriptions-c...

by kuu 5y ago

herodoturtle 5y ago

I assume you meant “pull” or “unplug” the cable :)

kuu 5y ago
Yes ;)

cma 5y ago

They'd rather refund small guys for mistakes than give big guys an easy limit to set.

kuu 5y ago
I guess big guys don't want they service to suddenly stop, so they probably would not use this... But it's just a guess
- electroly 5y ago
  Absolutely that. Storage costs money, so in order to absolutely cap your spending they would have to delete all your stored data, too. Deleting S3 buckets and EBS volumes on a spending blip is absolutely the last thing any company with any budget at all wants to happen, ever. It would be preferable for that not to even be possible in any situation. This is the sort of thing that only extremely small casual users want, and it isn't worth it to AWS to cater to those users. For everyone else, more complexity than a "kill everything at $X" switch is needed, and that's exactly what we do have. We don't get to absolutely cap our spending to the penny but we also don't risk having our data vanish because of a billing issue.
  - ghaff 5y ago
    I assume the sensible implementation would be cut off access and give you some period to settle your bill before the data is deleted.
  - CodesInChaos 5y ago
    > Storage costs money The dumb solution for that is to exclude persistent storage from the limit. The nice solution for that is supporting both "runrate" and "consumption" limits. Using a runrate limit, spinning up an instance, creating a file, etc. allocates budgets for running it continuously, which is released when shutting it down/deleting it. Hitting the limit prevents new resources from being allocated, but keeps existing ones alive. This should be used for persistent storage and instances used to handle base load. Using a consumption limit, the resource is shut down when the limit is hit. If the shut-off is delayed, the cloud service eats the overage, since they control the delay. This should be used for bandwidth, paid api-calls, and auto-scaling instances. The user should be able to create multiple limits of each kind, and assign different services to such limits. Alerts when going near the limit can help the user raise it, if that's their intention. For consumption, it might also make sense to have rate limiters, which throttle after a burst budget is exceeded, similar to how compute works on T instances on AWS. But those probably only make sense for individual services, not globally (e.g. throttle an instance to 100 Mbit/s after it exhausted its 5 TB/day bandwidth allocation, or throttle an API to x calls/s).
- cma 5y ago
  For background batch jobs and analytics etc. they might want caps. Say something like a video transcoding workload. And lots of things could benefit not from a cap, but some kind of gradual degradation in bandwidth/instance allocation + a warning so you can raise the limits, it doesn't have to just shut everything down immediately using a hard cap.

frameset 5y ago

But there is an option. In Azure you can "set a budget". He even goes over it in the post. Did you read the linked article?

lkxijlewlf 5y ago
https://news.ycombinator.com/newsguidelines.html https://news.ycombinator.com/newsguidelines.html

defaultname 5y ago

Oracle has fantastic budget tools. Not just "you've passed your budget", but "you're forecast to pass your budget in 22 days before the month is up". And you can couple it with quotas to create hard budgets. AWS has decent tools in this regard, but it pales compared to Oracle. Azure is a product I've never used with any scale (just small projects), but the fact that it actually costs money to setup alerts is gross (and morally reprehensible). Even if it's a trivial amount, that alone just sours the product in my eyes. I mean, already Azure is pretty uncompetitive unless you're running on free credits, as Troy apparently is (purportedly some $13K per year, so unsure what the pitch for donations to cover a bill is about).

schemescape 5y ago
This piqued my interest, but a few quick searches (using a search engine--the Oracle Cloud site search only turned up press releases...), indicate that quotas just prevent you from spinning up new instances. That's helpful, but I was hoping for some sort of way to cap my bill (for hobby projects), even if that requries deleting resources. Oracle Cloud has an enticing free tier, but I'm too afraid to use it because it requires a credit card and I don't see any way to put a monthly cap on my budget. (I'm sure hobby projects with ~$5 - 10/month budgets isn't their target market, but I can dream :) Edit to add the page I was reading: https://docs.oracle.com/en/cloud/get-started/subscriptions-cloud/mmocs/quota-breach-details.html https://docs.oracle.com/en/cloud/get-started/subscriptions-c...