We've been using CloudAnalyticsX for about a year, mostly on a pay-per-query model. It was predictable: we knew our average cost per 10k queries. Last month, our account rep offered a "free" upgrade to their new "TurboEngine." The sales pitch was all about performance gains with "no downside."
Fast forward to this month's invoice: our per-query costs have nearly tripled. The volume of queries didn't change significantly, and we didn't onboard any new users.
Has anyone else experienced this? I'm trying to reverse-engineer what happened. My theory is that the "free" upgrade changed how they *count* a query. For example:
- Previously, a single `SELECT` was one query.
- Now, with "TurboEngine," maybe each *stage* in a distributed query plan is counted separately?
- Or perhaps they've started charging for metadata queries that were previously free?
I pulled our query logs and compared patterns from before and after the upgrade. The structure is identical. I'm looking at the vendor's API response headers now, trying to see if there's a new `X-Billed-Units` field or something similar.
Key details:
- Vendor: CloudAnalyticsX
- Pricing Model: Advertised as "per-query"
- Change: "TurboEngine" infrastructure upgrade
- Result: 3x cost increase on stable workload
What should I be auditing in my logs or billing exports? I'm looking for:
* Changes in query execution time (are faster queries now more expensive?)
* New line items in the detailed billing CSV
* Any mention of "compute units" or "processing units" instead of "queries"
If you've been through this, what was the actual culprit? And more importantly, what's the best script or method to correlate our query logs with their billing line items to prove the discrepancy?