Chapter 17. Know Where to Scale
Lisa Huynh
Most of the time, if all goes well, an application will hit a point where it needs to grow. Outside of “the application is timing out,” however, determining an acceptable level of performance can be subjective. Someone whose customers are all in Canada may not care about response times in Japan.
Whatever your metrics, let’s say you’re there. Typically, we can upgrade our systems by scaling up or out, also called vertical and horizontal scaling, respectively. With vertical scaling, we upgrade a resource in our existing infrastructure. With horizontal scaling, we add more instances. But which one should you be doing?
Vertical Scaling
You’re hitting the CPU limit, so you upgrade your instance from one with 8 CPUs to 16. Or maybe you’re running out of storage space, so you go from 100 GiB to 500 GiB. With this easiest and simplest way to scale, you’re running the same application but on more powerful resources.
Most relational databases use vertical scaling so that they can guarantee data validity (atomicity, consistency, isolation, and durability, or ACID properties) and support transactions. So, for applications requiring a consistent view of the data, such as in banking, you’d typically stick to vertical scaling.
Unfortunately, this type of scaling often involves downtime. If there’s nowhere else ...
Become an O’Reilly member and get unlimited access to this title plus top books and audiobooks from O’Reilly and nearly 200 top publishers, thousands of courses curated by job role, 150+ live events each month,
and much more.
Read now
Unlock full access