5 things you have to know when you think about MartenDB

Most of my developer’s career I work with SQLServer as database. Mostly because for .NET it’s a natural choice and all of the tools are very well supported and integrated. But in my new project I want to use EventSourcing. After investigation I have found MartenDB. There are my first thoughts about it.

Types of data structures in MartenDB

The first thing which I notices is that there are three types of data structures in MartenDB.

• tables - the standard one, which I know from SqlServer
• documents - there is a place to store JSON data
• events - a place for events as a place for events in EventSourcing

It is really nice approach, because we can have one place to store all of these things in one place. I can imagine it will be much easier to prepare some read models or store data easly.

Automatic structure creation

When you use MartenDB we do not need to create data structure. It is done automatically. It is a fantastic feature. You can focus on the business logic. You just create a simple class and MartenDB will create a table for you on the fly. There are some benefits of that:

• You don’t have to care about migrations. At least the the beginning of the project.
• You can just think about storing a specific type and you can do it.
• Of course can modify the structure, if the default one is not enough for you. eg. you can add indexes.

I am no sure about the performance, but there are at least two pages in documentation how to deal with that. So I believe it is not a problem.

Events and projections

MartenDB is a tool for EventSourcing, so it is not a surprise that is has good support for events. When I was reading the documentation I was pleasantly surprised how easy it is to:

• store and load events
  • It is the base for EventSourcing. You need to store events. With MartenDB it is soo easy.
• create projections
  • Creating and refreshing projections and read models is a crucial part of EventSourcing. MartenDB makes it simple and natural

PostgreSQL used as a storage

It is a fantastic decision to use PostgreSQL as a backbone. I wrote in my previous article about Postgres in docker that I have no experience with PostgreSQL, but it is a good choice because:

• It is a standard tool which is common used around the whole world.
• There are huge amount of tools which can help you with Postgres.
• It is a very mature tool. It was created in 1996.

Documentation is great

The 1st thing I did with MartemDB was reading the documentation. It is easy to read and understand. It is full of examples. Just after reading the documentation I feel I am abe to work with that. Of course there are some more advanced topics I skipped or missed, like performance, but I know the information is there.

Conclusion

I will start migration my project with home-made event store to MartenDB. I believe that huge amount of can will be thrown away and you can be sure I will sate it with you.

What is you feeling about MartenDB? Let me know in comments.