Partitioned database q=1 single-node deployment: irreversible partitioning and compaction trade-offs for cost reduction

Evaluating a partitioned database with q=1 on a single node to reduce infrastructure costs for a low-traffic workload under 10 million documents. The goal is to eliminate cluster coordination overhead while maintaining query performance through partition-aligned access patterns.

Constraints include the irreversible nature of partitioned database creation, loss of global changes feed events for replication monitoring, requirement for manual compaction scheduling to avoid CPU spikes on small VMs, and the need to create Mango indexes explicitly for each new partition to prevent silent full-scan queries. Single-node deployment also removes automatic failover, shifting backup responsibility to external tooling.

• What monitoring strategies compensate for the absence of global changes feed events in partitioned databases?
• How does manual compaction scheduling interact with write patterns to avoid storage bloat on resource-constrained VMs?
• What partition-key design practices prevent silent full-scan queries when new partitions lack Mango indexes?