Lead Software Engineer at a tech vendor with 10,001+ employees
MSP
Top 10
Nov 1, 2025
I think Apache Kafka on Confluent Cloud can be improved by probably working more around Confluent or the tool. In my opinion, it should utilize the response structures in a better way or be able to detect if there is any variable or if there is any data structure that is mismatched, as it would be easier than us manually having to put in the exact name in order for it to match the response. Regarding additional improvements, I would say probably around error handling, where when we encounter errors specific to our response structures and everything, or the tables or anything of that nature, it would be better if we were prompted with better error handling mechanisms. I do not think there are any other improvements Apache Kafka on Confluent Cloud needs, aside from error handling and response structures.
If it were easier to configure clusters and had more straightforward configuration, high-level API abstraction in the APIs could improve it. The clustering is a little hard for juniors and clients. It's suitable for senior engineers, but the configuration and clustering are very hard for juniors.
Chief Architect at a financial services firm with 10,001+ employees
Real User
Top 20
Aug 19, 2025
In terms of improvements, observability and monitoring are areas that could be enhanced. They are lacking in terms of observability and monitoring compared to other products.
VP Engineering at a tech vendor with 1,001-5,000 employees
Real User
Top 10
May 28, 2025
I think what I would improve about the solution is the cost, mostly. From my standpoint, it's the cost. From an engineering perspective, it works really well. There's always room for improvement. One more point is sometimes it's more UI-related issues. Some of the more high-end features are more complicated to execute. But overall, it's a good product.
The ability to implement request-response communication on Apache Kafka ( /products/apache-kafka-reviews ) needs improvement. The schema registry is a bit misleading in terms of its location and how it works with multi-peer clusters.
Global Vice President, Product Strategy & Gtm at NucleusTeq
Real User
Top 10
Mar 4, 2025
Improvement can be made by making it easier to build applications on the real-time stream, focusing on real-time pre-processing and anomaly detection. They should enhance their capabilities in real-time data processing to support AI scenarios, in line with their messaging.
Learn what your peers think about Apache Kafka on Confluent Cloud. Get advice and tips from experienced pros sharing their opinions. Updated: January 2026.
Some areas for improvement in Apache Kafka on Confluent Cloud include issues faced during migration with Kubernetes pods. This aspect could be smoother to better support migration processes.
There's one thing that's a common use case, but I don't know why it's not covered in Kafka. When a message comes in, and another message with the same key arrives, the first version should be deleted automatically. We want to keep only one instance of a message at any given time, the latest one. However, Kafka doesn't have this functionality built-in. It keeps all the data, and we have to manually delete the older versions. So, I would like to have only one instance of messages, based on the keys. If the key is the same, there should always be the latest message present instead of all versions of that message.
The administration port could be more extensive. Additionally, managing the states of certain events could be made easier, perhaps with automatic rollback instead of having to program it manually.
Senior Architect at a outsourcing company with 501-1,000 employees
Real User
Nov 29, 2023
Regarding real-time data usage, there were challenges with CDC (Change Data Capture) integrations. Specifically, with PyTRAN, we encountered difficulties. We recommended using our on-premises Kaspersky as an alternative to PyTRAN for that specific use case due to issues with CDC store configuration and log reading challenges with the iton components.
For the original Kafka, there is room for improvement in terms of latency spikes and resource consumption. It consumes a lot of memory. Resource consumption. It consumes a lot of memory.
Apache Kafka on Confluent Cloud provides real-time data streaming with seamless integration, enhanced scalability, and efficient data processing, recognized for its real-time architecture, ease of use, and reliable multi-cloud operations while effectively managing large data volumes.Apache Kafka on Confluent Cloud is designed to handle large-scale data operations across different cloud environments. It supports real-time data streaming, crucial for applications in transaction processing,...
I think Apache Kafka on Confluent Cloud can be improved by probably working more around Confluent or the tool. In my opinion, it should utilize the response structures in a better way or be able to detect if there is any variable or if there is any data structure that is mismatched, as it would be easier than us manually having to put in the exact name in order for it to match the response. Regarding additional improvements, I would say probably around error handling, where when we encounter errors specific to our response structures and everything, or the tables or anything of that nature, it would be better if we were prompted with better error handling mechanisms. I do not think there are any other improvements Apache Kafka on Confluent Cloud needs, aside from error handling and response structures.
If it were easier to configure clusters and had more straightforward configuration, high-level API abstraction in the APIs could improve it. The clustering is a little hard for juniors and clients. It's suitable for senior engineers, but the configuration and clustering are very hard for juniors.
In terms of improvements, observability and monitoring are areas that could be enhanced. They are lacking in terms of observability and monitoring compared to other products.
I think what I would improve about the solution is the cost, mostly. From my standpoint, it's the cost. From an engineering perspective, it works really well. There's always room for improvement. One more point is sometimes it's more UI-related issues. Some of the more high-end features are more complicated to execute. But overall, it's a good product.
The ability to implement request-response communication on Apache Kafka ( /products/apache-kafka-reviews ) needs improvement. The schema registry is a bit misleading in terms of its location and how it works with multi-peer clusters.
Improvement can be made by making it easier to build applications on the real-time stream, focusing on real-time pre-processing and anomaly detection. They should enhance their capabilities in real-time data processing to support AI scenarios, in line with their messaging.
Some areas for improvement in Apache Kafka on Confluent Cloud include issues faced during migration with Kubernetes pods. This aspect could be smoother to better support migration processes.
The solution is expensive.
There's one thing that's a common use case, but I don't know why it's not covered in Kafka. When a message comes in, and another message with the same key arrives, the first version should be deleted automatically. We want to keep only one instance of a message at any given time, the latest one. However, Kafka doesn't have this functionality built-in. It keeps all the data, and we have to manually delete the older versions. So, I would like to have only one instance of messages, based on the keys. If the key is the same, there should always be the latest message present instead of all versions of that message.
The administration port could be more extensive. Additionally, managing the states of certain events could be made easier, perhaps with automatic rollback instead of having to program it manually.
I saw an interesting improvement related to the analytics environment.
Regarding real-time data usage, there were challenges with CDC (Change Data Capture) integrations. Specifically, with PyTRAN, we encountered difficulties. We recommended using our on-premises Kaspersky as an alternative to PyTRAN for that specific use case due to issues with CDC store configuration and log reading challenges with the iton components.
For the original Kafka, there is room for improvement in terms of latency spikes and resource consumption. It consumes a lot of memory. Resource consumption. It consumes a lot of memory.
There could be an in-built feature for data analysis.