Tracking Actual Project Costs for IDC-Hosted External Projects

Support
1

So for those researchers who are using IDC-provided projects as part of our credit program, you should be aware that Google just placed some credits in the billing account we use to cover those projects. It turns out that the project cost you see on your dashboard will be misleading, since the displayed costs are factoring in those cloud credits first. So you will probably see the cost as $0, which is misleading. Instead, go to the billing report page, and uncheck the boxes shown in the highlighted region below. That will remove the credit application and show you your true costs:

BillingChecksScreen
BillingChecksScreen2778×1455 331 KB

3 Likes
2

Thanks, we saw that and were wondering about that.

3

Yes, it’s going to be confusing for everybody. Just now seeing the effects of Google adding credits onto our cost tracking. So all your costs are covered by IDC, but until the credits are exhausted everything is going appear to be $0 following the April 7th infusion of credits. So if you want to learn about what you are spending, uncheck the boxes. And the number on your dashboard Billing tile is just going to be useless (I don’t know of a way to turn off promotional credits on that tile).

1 Like
4

We are working on a system that will report your true spending in your project logs. Note that what you see on your spending page is frequently 6-12 hours old. It is reported as Google prepares their billing for invoicing. We are working to have that logging report near real-time spending on VMs since that is the biggest concern for cost overruns. Stay tuned!

5

And thanks for keeping an eye on your external project spending!

6

thanks for the awesome information.

7

You’re welcome.

The realtime cost analysis system we are working on is making good progress recently. I’m frequently checking code into a branch of GitHub - ImagingDataCommons/RealtimeCosts: System for estimating realtime burn rates these days. In addition to ~realtime tracking of VM costs (now including sustained use discounts and the pricy GPUs), it is now also tracking ~realtime BigQuery costs, and estimating bucket storage costs based by extrapolating. Turns out that storage cost reporting can lag by 24 hours.