1. They have really good reporting options. You can see how much you're spending day-by-day and how much your estimated monthly total will be. You can break it down by service, or even by account with consolidated billing.
2. They have built-in billing alerts. You can get notified if your spend in any particular category is greater than some threshold.
3. We use a 3rd party service (Cloudability) to help with cost tracking and RI allocations. There are several other companies that do the same, and depending on your total AWS bill they have more or less automation available to you.
There is a caveat that applies for quite some of the audience here, I think: AWS credits.
If you are spending AWS credits all the billing reporting features break down, and it becomes quite hard to track down in which service you are burning your credits.
AWS' Billing dashboard is also great for a quick view. The report I use most is one I created that is the "Daily Cost per API Operation". Do the "Daily costs" report then group by API operation. I then view it by day (instead of month) for the last 3 cal months as a stacked bar graph.
I also have a CloudWatch alarm set to trigger at a certain spend. That alarm should trigger around the same day of each month. That way if it triggers 7 (or whatever) days before expected, I know we may have a problem.
Its been about 18 months since I was at an AWS-heavy shop. We used Netflix Ice (open source) to do cost reporting. We also heavily leaned on resource tagging to be able to report on segmentation - what teams were using resources, if it was dev, test, or prod etc. Tagging was crucial to managing costs for medium/large organizations and we spent a lot of time ensuring everyone was tagging their resources properly. I also made a one-liner change to Ice so that it would group untagged resources together so we could do reporting on how well we were tagging resources.
We tried ice but it felt it was clunky. Someone suggested just dump the detailed billing report with tags into tableau(or quicksite/Athena/etc) and graph the things you care about there, which worked well and we could apply transforms to the data easily. Also explored using config rules to enforce the tagging requirements. Tagging is really key to tracking this. We tested some of the third parties, and felt the only value was the ri adjusters, which we were able to work around
Cost Explorer as secondary or more detailed views.
We're mostly done migrating to a subaccount per squad (or even service) model, which means there's a useful report in cost explorer on the micro level. (We started in one massive, central, shared account and tried to get by with tagging; that was largely a failure on the tagging front, but it also made it quite difficult to ensure that you had all the access and freedom you needed without inadvertently having more access than you wanted to someone else's stuff.)
8 comments
[ 3.2 ms ] story [ 24.8 ms ] thread2. They have built-in billing alerts. You can get notified if your spend in any particular category is greater than some threshold.
3. We use a 3rd party service (Cloudability) to help with cost tracking and RI allocations. There are several other companies that do the same, and depending on your total AWS bill they have more or less automation available to you.
If you are spending AWS credits all the billing reporting features break down, and it becomes quite hard to track down in which service you are burning your credits.
AWS' Billing dashboard is also great for a quick view. The report I use most is one I created that is the "Daily Cost per API Operation". Do the "Daily costs" report then group by API operation. I then view it by day (instead of month) for the last 3 cal months as a stacked bar graph.
I also have a CloudWatch alarm set to trigger at a certain spend. That alarm should trigger around the same day of each month. That way if it triggers 7 (or whatever) days before expected, I know we may have a problem.
Cost Explorer as secondary or more detailed views.
We're mostly done migrating to a subaccount per squad (or even service) model, which means there's a useful report in cost explorer on the micro level. (We started in one massive, central, shared account and tried to get by with tagging; that was largely a failure on the tagging front, but it also made it quite difficult to ensure that you had all the access and freedom you needed without inadvertently having more access than you wanted to someone else's stuff.)