We have flink running over kafka for various aggregations. One of the streams we analyse is order-audits (Basically every state change is emitted as an event).
Each order-event is something like this
{
"id" : "ord-1",
"merchant_id" : "merchant-a",
"status" : "created",
...
"updated_at" :
"event_time" :
}
I want to run an aggregate which can aggregate at a merchant level for a given day.
Something like
{
"merchant_id" : "merchant-a",
"date" : "2019-07-01",
"started" : 10,
"completed" : 13,
"cancelled" : 3
}
Is flink a good fit for this type of aggregation? (Most of the examples are straight forward aggregations)
Sorry if this is repeated/naive. Thanks!
Sure, that kind of analysis is easily done with Flink. You'll probably find it easiest to do this with Flink's SQL API, as the learning curve there is gentle -- once you get setup, it's very straightforward, assuming you know some SQL.
Take a look at https://github.com/ververica/sql-training/ for a guided introduction.