Cookbooks vs manually setting up statsD/metric collection servers?

I am in the process of setting up a dedicated ec2 instance as a statsD server. I was wondering if there's a best practice around this. Allow me to elaborate. When dealing with cloud infrastructure, I found Terraform to be extremely useful. All the infrastructure you need is expressed and code and any changes to this code base of terraform modules can be tracked effectively. It also makes sense to have it in the same repository as your source code. So whenever CD kicks in, we can make sure our infrastructure is updated as and when needed.

I have a similar question around a statsd server. I came across the Chef Configuration Management tool but given the size of our operation at this stage - it feels like an overkill. I am curious to know what do people do for such servers. Do they prefer manually managing these? Or is there a way to express it as code - like chef. Or probably something else I don't know about.

Solution

At the risk of sounding opinionated, I would say that maintaining configuration using configuration management tool such as Chef is a good idea regardless of the size of the infrastructure.

However for your situation, you should evaluate below points:

Although your current requirement is one statsd server now, do you foresee requirement for additional machines?
Configuration management tools like Chef are components of your infrastructure, which need to setup as well. Is it feasible for your current requirement or something in near future?
In most cases you will be able to reuse community work, such as statsd cookbook from Chef supermarket. If putting effort into automating it yourself is your concern.