The viability of monoliths

Cloud services like AWS and Microsoft Azure are very popular nowadays and have almost completely taken over hosting. The complete set of tools and services they offer with the added bonus of high flexiblity in hosting offers a lot of benefits, but is it really the best solution for every use case? I would like to argue that the answer is no

Over architecting

Using multiple instances instead of one big instance is mostly because of the popularity of microservices. The idea of using microservices is great in splitting the logic of an application vertically instead of horizontally, which means that each feature and corresponding team maintains the whole chain from database to the client, ideally at least. In the past this used to be more split up. I think using microservices is great but I think it might be overdone a little. It introduces a lot of layers of complexity and most of the time every team gets to choose what stack they use. This often causes an application to become very cluttered with many different kinds of services using different technologies which depend on each other. There are even jokes about it nowadays, that to get a birthday from a customer you have to go through multiple services just to get this simple date displayed so the estimated time to develop this feature is over one year. This is an exageration of course but these kinds of problems are taking longer than necessary in a lot of cases.

I think central services within an application should be managed by one big system, for instance handling database operations like a read or write, authentication and other basic operations.

Another downside of this split is the added cost of running all these services atomically. Every service has to be managed and monitored.

Over scalability

I think the scalability that cloud providers offer is great; but how often do you really peak above mean load or have very low traffic on your platform. I think a combination of one big dedicated server which gets help from scaled instances can also offer this advantage. The advantage of this big dedicated server to handle mean load is that it is cheaper and you will get more control over your own server. Virtualising still runs on top of a hypervisor which a dedicated server does not need. This construction makes it so software that need kernel level drivers that are not there can not be installed, while they can on dedicated servers

Over specialisation

The microservices landscape where each team works isolated on a specific number of services also reduces communication between different teams. Teams will be working on more isolated parts of the application. I think communication between teams is very important to not only have a more healthy social environment in a company but also to have more creativity that can be used to solve complex problems. Maybe a developer in another team has already fixed a similar problem that your team has been struggling with for weeks.

Over complication

Using a scalable solution where the amount of instances can be scaled up and down also adds a new layer of complexity, which for a lot of companies is definetely needed, but for some is quite overkill. For a large scale organisation using kubernetes or similar tools is a must! But is it really necessary to run smaller projects in it, just a simple docker container using docker compose is enough for a lot of use cases in my opinion. I think we as developers get too drawn to feeling smart and wanting the best, most technically amazing solution, but sometimes the problem that has to be solved just does not require this big of a solution.

Over reliance

The best pricing and features are currently only available with the biggest companies right now, these are Google, Microsoft Azure and AWS. Where AWS and Microsoft Azure together control about 50% of this market. This raises concerns for me since this control of the market means that these companies have a lot of power. Sadly only one of these things can be solved by using a more monolithic approach which is the price. The amount of features you will have to develop yourself which is still a big obstacle. I do not have a solution for this problem, but I am concerned what this growing power of these companies will do.

Conclusion

After reading the above you will probably get the idea that I am not a fan of all these modern auto-scaling and orhcestration tools, but on the contrary! I think this technology is amazing and have also tried Google Cloud, Microsoft Azure and AWS. I think it is good that these services and their competitors have changed the landscape of how servers run. My criticism is that you should not overdo it. I think it is not the perfect solution for every problem.

I think reducing some of the servers to a dedicated server which can handle mean load, can reduce some of the complexity of a system and reduce running costs. If you don't believe me even companies like Amazon for hosting Amazon Prime has shifted to doing this, YES the same Amazon that tries to sell us AWS.