# Producer Sender Metrics : request-size-avg does not scale with # of partitions

**URL:** <https://forum.confluent.io/t/producer-sender-metrics-request-size-avg-does-not-scale-with-of-partitions/2998>\
**Category:** Clients\
**Created:** [5 October 2021 15:28 UTC](https://forum.confluent.io/t/producer-sender-metrics-request-size-avg-does-not-scale-with-of-partitions/2998 "2021-10-05T15:28:24Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![gautamg](https://avatars.discourse-cdn.com/v4/letter/g/76d3ee/32.png) [@gautamg](https://forum.confluent.io/u/gautamg)\
**Post date:** [5 October 2021 15:28 UTC](https://forum.confluent.io/t/producer-sender-metrics-request-size-avg-does-not-scale-with-of-partitions/2998/1 "2021-10-05T15:28:24Z")

</div>

I am running the performance test with the below configurations

- batch.size = 16,384. (default batch size)
- No. of partitions = 3

In the metrics, I see that

- request-size-avg=16285.006 (same as batch size)

With the # of partitions being 3, I would expect it to be 16,384 \* 3 = 49,152 bytes.

What could be the reason behind this behavior? Does kafka further optimize batching that is not very transparent when it comes to batching?

I see the same behavior where the request-size-avg is same as batch.size even when I increase or decrease the batch size from the default number.

Thanks.
