Startup founders have no end of infrastructure hosting choices. Making the wrong choice can limit agility, increase costs, and put a brake on growth. Technological debt applies to infrastructure too, and infrastructure decisions made in the early days of a company may well cause headaches down the line.
There are any number of ways we might compare infrastructure options, but today I’d like to consider the issue by focusing on the two most popular alternatives: bare metal or dedicated servers, and cloud servers.
Each is potentially a good choice, and there are plenty of startup success stories involving companies that opted for cloud and dedicated platforms.
However, there are differences, and it’s worth understanding how those differences might impact your startup.
Just so we’re all on the same page: a cloud platform offers virtualized servers that can be deployed instantly, are controlled by an API or web interface, and that offer metered billing.
Dedicated servers are physical servers: a computer with a processor, RAM, and storage. They live in a rack in a data center and the user has complete access to all the resources and capabilities of that machine. Dedicated servers are usually billed over longer periods than cloud servers, often monthly or yearly.
Which is better for startups? It really all depends on the specific situation, but let’s take the example of an early stage startup creating a SaaS application. Our example startup has raised funds from angel investors, but has yet to receive a significant injection of VC funds. What are the infrastructure requirements of this company?
For this startup, a cloud platform is almost certainly the best choice. Development, testing, and production servers can quickly be spun up and discarded when no longer needed, allowing the company to quickly iterate and get code into production. The company will only ever pay for the resources it uses.
However, there might be occasions where dedicated hardware is more appropriate. If performance is of particular concern — heavy duty data analytics, machine learning over large datasets, fin-tech applications where low latencies are a priority — choosing dedicated servers makes sense.
Dedicated servers are also a great choice for long-running applications where elasticity is less of a concern. You might choose to host a critical database on a dedicated server and use cloud nodes for scalable application and web servers that access the database.
It should be kept in mind that the decision isn’t really between cloud and dedicated, but which combination of infrastructure types best suits the specific needs of the business.
In a future article, I’ll discuss another big infrastructure choice that faces startup founders: opting for managed or unmanaged infrastructure.