[Roadmap] Planning 0.8 Release

[wuchong]

Timeline Plan

• Feature Freeze: August 30, 2025
• Release: September 25, 2025

*The v0.8 will be the first release in the Apache incubator. We left nearly a month to resolve all the issues. *

Milestone Board

https://github.com/alibaba/fluss/milestone/4

Highlight Features

List of features announced by contributors and committers that are likely to be ready for the feature freeze

Streaming Lakehouse Production-Ready

• Union read should be supported in streaming mode #1140
• [Feature] Log scanner should support fetch the logs tiered into lake #41
• [FIP-6] Union read should be decouple with Paimon #1141

Streaming Lakehouse for Iceberg

• Support Iceberg as Lakehouse Storage #452

Streaming Lakehouse for Lance

• [Umbrella] Support Streaming Lakehouse for Lance table format #1155

Full DataType Support & Schema Evolution

• [Umbrella][Feature] Support for complex data types #816
• [Umbrella][Feature] Schema Evolution Support #1154

Cluster Operations

• Rolling Upgrade: Support ControlledShutdown for TabletServer #1142
• Replica Rebalance: Fluss cluster supports rebalancing #1144
• Throttle: Support cluster throttling capabilities #1145
• Coordinator HA: [Feature] CoordinatorServer Supports High-Available #188

Stabilities

• Refactor metadata cache and async updating mechanism in client side #483
• Use Leader Epoch rather than High Watermark for Truncation #673
• Remove TableInfo from server MetadataCache #982
• [Umbrella] Add backward and forward compatibility E2E tests #1161

Performance

• [Feature] Support filter pushdown for Log Tables #197
• Optimize auto drop partition perfomance #864

Flink Connector

• Support Flink 2.0 & 2.1: Support Flink 2.0 #651
• Support Delta Join: Support DeltaJoin on Flink 2.1 #1143
• Support Materialized Table: [Feature] Support for Flink materialized table #860
• Support $changelog system table: Support $changelog auxiliary table for flink connector #356
• Support $binlog system table: Add binglog system table support for fluss tables #978

Spark Connector

• [Umbrella][Feature] Support spark engine #155

Adds native support for Spark as a computation engine, enabling efficient batch and streaming and interactive query processing.

Herm Charts

• Add a helm chart for Fluss #779

Easier deployment for users and increase adoption.

Python Client

Support Python SDK to connect with Python ecosystems, such as PyArrow, Pandas, Lance, DuckDB.
This may happen in a separate repository, e.g., apache/fluss-python.

Filesystems

• GFS: Implement obtainSecurityToken for google filesystem #993
• MiniO: [File system] Add AssumeRoleWithWebIdentity STS functionality for temporary credentials to S3 file system to support MinIO #621

Others

• Support JDK11: Upgrade default compile language to use Java 11 #1156
• [metrics] OpenTelemetry metric reporter #181

---

We welcome any ideas, suggestions, or feedback from the community! Your input is invaluable in shaping the future of Fluss. Please feel free to comment on the discussion and linked issues.

[polyzos]

What about the Python client?

[SteNicholas]

What about the Python client?

What about the Restful API via openapi？

[michaelkoepf]

@SteNicholas maybe we can have a more in-depth discussion about this in the next meeting. in particular regarding the choice of framework to be used (spring boot, quarkus, ...)

[MehulBatra]

Should we account for potential changes that may be required for the project after entering incubation as part of this plan?

[wuchong]

That should be included in v0.8 milestone, but it's not user-facing feature, so I didn't list them here.

[MehulBatra]

Makes sense, thank you!

[MehulBatra]

Should we include support for ChangeLog and Binlog auxiliary tables via the Flink Connector in v0.8?

[michaelkoepf]

• minio support: [File system] Add AssumeRoleWithWebIdentity STS functionality for temporary credentials to S3 file system to support MinIO #621
• otel collector: [metrics] OpenTelemetry metric reporter #181

the best case would be to have helm chart, minio and otel collector in one release, so fluss can be considered "kubernetes/cloud-native-ready"

[polyzos]

It would be awesome to have these as well

[rafaelsmf]

Hey guys,

I’d love to contribute to these features, as I’m planning to start a POC with Fluss and need to deploy it in our Kubernetes cluster to integrate with other systems on AWS. So, if you’ve already started working on this or plan to do so, I’m happy to help as well.

[michaelkoepf]

another thing which i already added to the agenda for the next meeting: i think we also need a regression test suite for backward compatibility -- rather sooner than later. maybe somebody from the existing contributors knows somebody who has experience with this, or is at least interested in contributing it.

[wuchong]

+1, that's planned. I will create issues for that.

[georgelza]

don't see comment to Apache Kafka Client/protocol for 0.8.0

[polyzos]

@georgelza Kafka compatibility is postponed for now to prioritise the above features

[rafaelsmf]

I’d suggest adding a troubleshooting page with common issues and how to resolve them when using Fluss.
For example: if my application goes down, how can I recover it?

[polyzos]

Thanks for the feedback.. This should definitely be useful for users.

[jackylee-ch]

I noticed that the code still depends on com.alibaba.fluss. Could we switch it to org.apache.fluss in version 0.8?

[polyzos]

Yes it will be switched