Researching the right tool (vs Apache Flink?)

EventStoreDB
1

Hi guys!

We are bringing an event processing system into our platform to react better to overall system and user behavior. Until we found Event Store today, we were sold on Apache Flink which is a proved stream processor with CEP on top.

Now I am confused, does anyone have experience with both of them? Any insight is welcome at this point. :slight_smile:

(performance wouldn’t be a problem, our data is not too big)

Thank you!

Oytun

2

Hi Oytun,

Flink is part of a large ecosystem and you need to combine it with other tools. This can be powerful or be an administrative overhead depending on the available skills.

BTW whenever I found myself in that sort of situation I end up implementing a simple spike solution to help make the decision.

Hope this helps

Riccardo

3

Hey Riccardo!

Thanks for replying. Definitely, a spike test should let us see how we feel with each implementation. I guess my current parameters are:

  • Ecosystem / community support (both Flink and ES have amazing community, like you!)
  • Professional support (Flink doesn’t have official business support, but a company, dataArtisans, focuses on Flink)
  • Load handling / battle tested (Flink is definitely battle tested, ES looks like that, too, but not too many enterprises are using ES at the moment)
  • Integrations (Flink is heavily based on extensions to consume or publish (source/sink), our event bus, RabbitMQ, has an integration with Flink)
  • CEP power (Flink feels more powerful in this sense, having more verbs and expressions ready to use)
    Let’s see what’s going to come up with these in mind!

Oytun

4

"

  • Integrations (Flink is heavily based on extensions to consume or publish (source/sink), our event bus, RabbitMQ, has an integration with Flink
    What would you want in an integration? Just getting the messages in as example would be an hour’s work at the most. How would you see integration working beyond just getting messages in/out?
5

Hi Greg,

Nothing more than that actually. It’s not that integrations are difficult to create in ES, but why develop when there is an existing integration which is used by people in the ecosystem. Especially when you want to quickly use multiple connectors for different streams etc, they come in handy. When the software provides these integrations in their doc, it feels more healthy ( == perhaps more natively integrated).

So the integration article is more of a convenience.