Scalability in
software development is the ability of the system to adapt to increased demands, whether in terms of performance, maintenance, availability ...
Asking whether the system is scalable, can also be put as “will it stand the test of time''. As such, making a scalable system is the ideal goal under equal conditions. Too often, however, businesses give priority to costs over scalability, or even completely discard its importance.
Types of Scalability
Scalability of Performance
High performance is essential for the support of a lot of users. Preferably the system should serve 1 user as it would 100, 10 000 and so on. Here we differentiate two subtypes:
- Vertical scaling - it involves growing by utilizing more advanced or more powerful hardware.
Scaling vertically grants better performance than scaling horizontally. Everything is in one place, allowing faster returns and less vulnerability.
The problem with this approach is that there is only so much room to grow. At some point, we get diminishing returns as hardware gets more expensive. Also, the implementation of new hardware takes time.
- Horizontal scaling - is used much more widely for enterprise purposes.
Scaling out, software grows not by using more advanced or powerful hardware, but by adding more of it and spreading the workload throughout the new infrastructure.
This approach costs less and is easier to implement, and as a result, it is considered a better solution.
Scalability of Availability
How well the system responds to a wide area distribution. The CAP theorem says that a distributed system can not guarantee the reliability, availability, and partition tolerance at the same time.
Strong consistency assurances are often dropped in favor of the other two in a number of web applications. Rather, they depend on eventual consistency: the system will allow all users to read the latest updates at some point, but not all of them will do so immediately. This significantly reduces the synchronism of the application and allows the system to remain available to all users, even in the face of network partitions. Users may see inconsistent data during the partition, but the system’s consistency will eventually be restored.
Scalability of Maintenance
An easily overlooked aspect of the initial design phase, it represents how much more maintenance is required as the system scales in other aspects. This should preferably be less-than-linear, but it's the opposite all too often. It should be automated as much as possible.
Feature scalability
It’s a flexibility of system design to accommodate new features. A codebase that needs to be refactored for each new feature would have very low scalability, while a completely no-touch plug-in architecture would have very high feature scalability.
Scalability of Expenditures
The total cost of ownership includes development, maintenance and service expenditure. When designing a system, you need to balance the use of existing components and develop everything yourself.
Having industry-standard software could make it easier to employ an expert in the future.
Summary
Prioritization of scalability prepares your
company for success. Consider it early, and when it is most needed, you can reap the benefits in agility.
