A SaaS company hosts 60 separate Azure SQL databases, one per customer. Each database has unpredictable, sporadic activity: typically only 5-8 databases are busy at any given moment, while the rest are nearly idle, and peak times differ per customer. The finance team wants to reduce the cost of provisioning dedicated compute for every database while still allowing any individual database to burst to higher performance when its customer becomes active. Which deployment approach should you recommend?