Vertical scalability
Also called scaling up.
Vertical scalability is a system's ability to handle more work by giving one server or machine more resources, such as CPU, memory or storage, rather than adding machines to share the workload.
How it works
You resize a virtual machine, add RAM to a physical server, or move a workload to a more powerful machine. The application often needs fewer changes than it would to run across several servers.
The upgrade must address the actual bottleneck. More memory helps a memory-constrained database; it will not fix an overloaded network connection.
There are limits: a machine has a maximum size, upgrades can require downtime, and a larger server does not by itself remove a single point of failure.
Doubling server resources without doubling performance
A startup runs its application on one virtual server. It considers this hypothetical upgrade; the prices are illustrative, not a provider quote.
| Resource | Current server | Upgraded server |
|---|---|---|
| Virtual CPUs | 4 | 8 |
| RAM | 16 GB | 32 GB |
| Monthly server cost | $80 | $160 |
CPU and RAM both double. The monthly bill increases by $160 − $80 = $80, or 100%.
That does not guarantee twice as many requests per second. If the application spends most of its time waiting for a separate database, upgrading the application server might achieve little. Compare performance before and after at the same workload.
For an early-stage startup, scaling up can postpone a complex redesign. The useful question is whether the extra capacity buys enough time to justify the recurring cost.
Questions people ask
What is vertical vs horizontal scalability?
Vertical scalability means increasing the capacity of individual machines. Horizontal scalability means adding machines that share the work. Upgrading one database server is vertical scaling; adding application servers behind a load balancer is horizontal scaling. Horizontal scaling requires software that can distribute work, and the two approaches can be combined.
