Ah, yeah, you weren't the only one bitten by that. We actually went and changed the Cloudant metering model recently so that you're billed on provisioned throughput rather than total request volume. You get dramatically more predictable billing, with the tradeoff that clients need to handle a 429 Too Many Requests response if the Cloudant instance is under-sized. More here:

https://www.ibm.com/blogs/bluemix/2016/09/new-cloudant-lite-...