What is vertical scaling in distributed databases?

Vertical scaling in distributed databases refers to increasing the capacity of individual nodes within the database system, such as adding more CPU, memory, storage, or other resources to a single server. This approach focuses on enhancing the performance of existing machines rather than expanding the number of nodes in the cluster. For example, if a database node is struggling to handle increased query load, upgrading its CPU from 8 cores to 16 cores or expanding its RAM from 64GB to 256GB would be vertical scaling. This method is often simpler to implement because it doesn’t require changes to the database’s architecture or data distribution logic. However, it has inherent limits, as hardware upgrades can only go so far before hitting physical or cost constraints.

A common use case for vertical scaling is when a workload grows predictably and doesn’t require immediate horizontal expansion. For instance, a distributed database powering an analytics platform might experience gradual increases in data volume. Upgrading individual nodes’ storage or processing power could temporarily address performance bottlenecks without reconfiguring sharding or replication strategies. Another example is legacy systems that weren’t designed for horizontal scaling—vertical scaling allows incremental improvements while minimizing operational complexity. However, this approach works best when the workload isn’t highly distributed or when specific nodes handle isolated tasks (e.g., a primary node managing writes in a leader-follower setup).

The main drawbacks of vertical scaling in distributed systems are cost and scalability limits. High-end hardware becomes exponentially expensive, and there’s a ceiling to how much a single machine can handle. For instance, a node with 2TB of RAM might cost significantly more than spreading the same workload across ten smaller nodes. Additionally, vertical scaling can introduce downtime during upgrades, which conflicts with the high-availability goals of distributed databases. While vertical scaling is a quick fix for short-term needs, most distributed databases prioritize horizontal scaling (adding nodes) for long-term scalability. That said, vertical scaling remains useful in hybrid scenarios—for example, boosting a coordinator node’s capacity to manage metadata more efficiently while relying on horizontal scaling for data storage.