Scalability is not (only) about computational power
When it comes to scalability, most people think that it's all about HA systems, oversized machines (or cloud instances) etc. In general, they just think about the whole scalability issue in technical manner.
Well, it is technical but from my point of view scalability is more about how you can scale as the whole company and company is not created just from computers, but also people.
Company pays money to programmers, therefore it is crucial for any company to use programmer's time as effectively as possible. It is crucial for any company to increase the long-term value of the whole company (or its product) while using the human power of programmers.
How long does it take a new programmer to understand your codebase in order to make changes effectively?
Does it take weeks, months, or even years? You're probably doing good job if it takes just days.
Programmers usually spend more time on reading (understanding) the code then writing it, that's just the fact, but you can still decrease that amount of time needed to understanding the code.
This is not a definitive list, but it is the list of my recommendations coming from few years of my programming experience:
There's just no better way of getting know each other's code than doing code reviews. There's much more (advantages) behind code reviews, which would be for separate article.
In fact, two persons (author + reviewer) knowing any specific piece of code is always better than just one (author). Ideally, you can even find a bug or performance issue before it even reaches any user testing phase.
And - what's more important - when there are two different opinions (between author & reviewer), these guys will have to find compromise, which will most probably help any third guy to understand the code easily and it will help both guys to learn from each other.
2. Define Coding Standards
Coding standards doesn't need to be any sign of large corporation or European Union, where you need everything to be standardised. It may help all members of your team, even you and even the future members.
It actually helps you to become one team instead of bunch of individuals. It is obviously important to find a consensus and ensure that the majority of your team agree with the definition of standards and again, it is not necessary to standardise everything.
You can start with just couple rules, eg. where to put braces, how many (if any) spaces to put around which keyword etc.
When you write down this list, it is also good to have tools in place which will help all developers easily check whether the new code is compliant with these standards.
3. Choose a “good” framework
At first, it is really useful to use an open-source framework. Unless you're Google or someone, who anyway probably already tried to use all available open-source solutions, only then you may think about building your own framework (actually you should have a really good reason even if you're in that situation).
Why? You save time (= money) and headaches of your development team. Why? Because someone else in the community already covered the most use-cases. Why? Because you can scale up by just using existing code somewhere in the community. Why? Ehh...
Why on earth would you like to spend money on reinventing the wheel? There's probably no reason, but still so many companies are reinventing the wheel (or actually trying to do so).
What's the conclusion? Choose a framework which is well extendible and have active community.
This would be for a whole new article or even more articles, but to make the long story short, just read the Clean Code.
Read it and encourage all other members of your team to do so as well.
This may sounds like an advertising, but this book has been written by programmers for programmers. These guys created its own best-practices which helped them to increase readability of code = reduce the amount of time needed to understand the code by others = save money.
Be aware that others === yourself after couple months.
You would probably find the same things after couple years of coding, but here you can have all this available without having to make your own path with many fails on it.
