What are the drawbacks of using Oracle databases?

Oracle Database costs more to license and run than most competing engines, and that cost is not fixed. Oracle licensing is based on the number of processor cores, adjusted by a per-processor core factor, so the bill grows every time you add CPU capacity rather than staying flat as usage changes.

When are those drawbacks worth accepting?

Large organizations running transactional systems with strict uptime requirements often keep Oracle despite the price. Real Application Clusters let a database run across multiple nodes at once, so one server failing does not take the system offline. Banks, insurers, and airlines lean on this for order processing, claims handling, and reservations, where an outage has a direct financial cost.

Oracle also holds up well under audit. Its ACID compliance and mature backup and recovery tooling satisfy regulated industries that need a documented, provable data integrity story. If a company already has years of PL/SQL logic in production, the licensing bill can still work out cheaper than rewriting that logic for another engine.

When do the drawbacks outweigh the benefits?

Should a new project default to Oracle? Usually not. Core-based licensing makes it one of the more expensive database engines to operate at scale, since cost tracks CPU count rather than data volume or query load. A team running a modest transactional workload can end up paying enterprise rates for capacity it does not need.

Administration is a second cost. Oracle deployments typically require a certified DBA to manage tuning, patching, and backup strategy, work that PostgreSQL or MySQL setups can often handle with a smaller team for the same workload size.

Is vendor lock-in an exaggeration? Not really. Oracle's PL/SQL dialect and proprietary features, including its partitioning options and optimizer hints, do not carry over to PostgreSQL or MySQL. Migrating later means rewriting stored procedures, not just moving data across.

Practical note

Before committing to Oracle for a new system, model core-based licensing costs against open-source alternatives at your expected scale, and check whether RAC-level high availability is an actual requirement or just a safety margin nobody has tested against real failure scenarios.