Today I would like to share my view on building up IT infrastructure for a small company. It’s important to do it right in this case.
When it comes to IT infrastructure, there is some good news for you. There are plenty of qualified suppliers out there. They will help you to avoid useless waste of money and issues.
I think it’s more important to have some flexibility and extensibility when we talk about small companies. You also need to pay only for the options you are really going to use.
There are some reasons why you don’t need to build infrastructure with servers and administrators:
It’s simply expensive. You will need a pull of servers in order to organize fault tolerant work. The usage of such resource is going to be low at the beginning;
Scaling such infrastructure demands strict planning in terms of future requirements growth which is not always possible;
You will need an OPS team to support this, which is not so bad but rather expensive (and you are going to face the problem of making people stay. We are going to discuss it later);
You will have to build the whole fault tolerant cycle that has been already created by suppliers. You will also need to arrange OPS team 24-hours shifts (and that is expensive).
What's the solution?
So, how can you save yourself some money on infrastructure without loosing flexibility and scaling? Use cloud hostings (such as Gae, Amazon, DigitalOcean…). Ideally don’t choose a virtual machine you will need to assemble. Take a Kubernetes cluster (you may use it for creating almost any solution without thinking of programming languages) and cloud database (again, you may choose one of the many out there). As a result you are going to get a cluster configuration and a database with clear monitoring and SLA. Some third-party companies take responsibility for scaling and fault tolerance. Any of such companies may encounter some failures, however it happens much less often in comparison with your own infrastructures usage (especially at some early stages). You will also need people to configure everything. As the time goes by, the needed amount of such specialists is going to be less and their expertise won’t be so unique. At this time you won’t need a classic OPS (and I am talking about at least two people you don’t need to hire). You will only need a DevOps or a developer with some understanding of cluster and database configuration. If you need help with choosing the right cloud solution, drop us a line.
If you have any questions or comments, you can use the feedback form to contact us: