# Ordered Processing of Messages consumed from Kafka
**URL:** https://discuss.akka.io/t/ordered-processing-of-messages-consumed-from-kafka/4178
**Category:** Akka Streams & Alpakka
**Tags:** kafka
**Created:** [May 12, 2019, 5:55pm UTC](https://discuss.akka.io/t/ordered-processing-of-messages-consumed-from-kafka/4178 "2019-05-12T17:55:18Z")
**Posts on this page:** 2
**Page:** 1
### Author:  [@sigurd](https://discuss.akka.io/u/sigurd)
#### Post date: [May 12, 2019, 5:55pm UTC](https://discuss.akka.io/t/ordered-processing-of-messages-consumed-from-kafka/4178/1 "2019-05-12T17:55:18Z")
Hey guys,
I have a question about the committing of messages in Kafka.
Kafka use a offset which is represented by a ascending number for a consumer group.
So the committing have to be done in correct order, right?
If I have three messages a,b,c and in this case a batch size of 1, then a need to commit a, then b and finish with c.
It is allowed to use a unordered operator such as mapAsyncUnordered?
In this example, the third line is the important part (process messages of one partition):
```auto
committablePartionedSource(..)
.mapAsyncUnordered(bufferSizer, pair -> pair.second()
.mapAsyncUnordered(buffer, business())
.runWith(commiterSink(),mat))
.toMat(Sink.ignore(), Keep.both())
.mapMaterializedValue(pair -> Consumer.createDrainingControl(pair))
.run(mat)
```
Is this process correct or break this the committing process of kafka?
Best
Sigurd
---
### Author:  [@johanandren](https://discuss.akka.io/u/johanandren)
#### Post date: [May 13, 2019, 9:10am UTC](https://discuss.akka.io/t/ordered-processing-of-messages-consumed-from-kafka/4178/2 "2019-05-13T09:10:04Z")
If you want to keep the ordering of the events you cannot use `mapAsyncUnordered` but must instead use `mapAsync` which keeps the original ordering.
With re-ordering and committing offset to Kafka you could end up committing a higher offset and then dropping a lower offset element and losing data.