Apache Incubator Case Studies

This page compiles case studies of podlings that have undergone the Apache Incubator process.

Each example shows different outcomes: healthy graduations, rapid success, recoveries, stagnation, retirement, withdrawal, and unique challenges such as vendor dominance or global contributor bases.

The goal is to help mentors, podlings, and the Incubator PMC learn from past experiences.

Note this is accurate as of October 2025.

Synthesis: Patterns Across Case Studies

What success tends to look like

• Diverse contributors beyond the origin company (new committers keep arriving).
• Steady ASF-compliant release cadence (and rotating release managers).
• Decisions and design resolved on dev@, not in private or vendor channels.
• Clear branding and governance neutrality; ecosystem growth invites more organizations in.
• Documentation and outreach that include global, time-zone, and language considerations.

Common failure or retirement signals

• Single-vendor dominance (most commits, decisions, or infrastructure controlled by one org).
• Irregular or stalled releases; few people able or willing to RM.
• Off-list decision-making; GitHub-only culture with thin mailing-list history.
• Shrinking participation; little path for newcomers to become committers or PPMC members.
• “Fit” mismatch with ASF processes (e.g., never shipping an incubating release).

Edge outcomes to note

• Withdrawn - Strong external ecosystem but misalignment with ASF processes; leaving can be the right call.
• Graduate then Retire - Some projects meet graduation criteria but fade soon after. Graduation confirms process maturity, not guaranteed longevity. Without fresh contributors, user engagement, or technical evolution, even a healthy TLP can decline into inactivity.

Additional Takeaways for Mentors

1. Early alignment matters
Projects that thrive (for example, Flink, Iceberg, and Airflow) had mentors who clarified ASF expectations early, especially around releases, mailing-list use, and decision-making. Early reinforcement prevents years of off-list or GitHub-only habits that are hard to undo later.

2. Balance guidance with independence
Over-involvement can mask weak community ownership. Mentors succeed when they teach why ASF practices exist, then step back and let the PPMC lead, intervening only when culture or process drifts.

3. Watch for silent warning signs
Long gaps between releases, single-company commit patterns, or mailing lists that go quiet are all early indicators of risk. Mentors should surface these gently before they become structural problems.

4. Model transparent communication
When mentors consistently reply on-list, summarize outcomes, and reference policy links, podlings mirror that behavior. Mentor silence or private side-chats often lead to slower culture adoption.

5. Recognize when withdrawal or retirement is healthy
As seen with Zipkin or Mesos, not every project needs to remain at Apache forever. Helping a community make that decision transparently can be as valuable as guiding one to graduation.

6. Keep mentorship time-bounded
Effective mentors plan their exit. Ensuring the PPMC can self-govern before graduation and remains functional without external prompting.

7. Time in incubation is a signal, not a goal
Incubation length varies from months to years. Fast graduation can show strong existing maturity (as with Flink) but may hide dependence if mentors push too soon. Longer incubation can simply reflect complex IP clearance (NetBeans) or gradual culture-building (Superset). Mentors should judge progress by ASF maturity, including steady releases, independent decisions, and visible community growth, not by time alone.

8. Mentorship style should adapt to the podling’s phase
Early phases need hands-on clarification of ASF basics such as releases, voting, and mailing lists. Later stages benefit more from coaching and accountability than instruction. Mentors who shift from directing to advising help the PPMC gain true independence before graduation.

9. Maintain context after graduation
Graduation isn't the end of learning. A short transition period where mentors remain subscribed or available helps the new TLP stabilize and handle first-year governance and reporting challenges confidently.

10. Reflect and share lessons back to the IPMC
Each incubation reveals new edge cases in culture, licensing, or governance. Mentors strengthen the whole Incubator when they share anonymized insights that inform future guidance and policy updates.

Case Study 1: Apache Pinot - Strong Community Growth and Graduation

Entry

• Created at LinkedIn in 2013 and open-sourced in 2015.
• Entered the Apache Incubator in October 2018.

During Incubation

• Releases: First Apache release in 2019, followed by a steady cadence through 2020.
• Community growth: Attracted contributors from Uber, StarTree, and independent volunteers beyond LinkedIn.
• Process adoption: Over time, discussions increasingly moved from GitHub and Slack to the dev@ mailing list to align with ASF norms.
• Mentorship: Mentors emphasized mailing-list transparency and release discipline.
• Adoption: Widely used in industry (LinkedIn, Uber, Stripe, Walmart, DoorDash).

Graduation

Graduated as a Top-Level Project in May 2021.
Current Status: TLP

Takeaway

Pinot shows that podlings with strong corporate origins can graduate successfully if they:

• Attract contributors from outside the founding company
• Deliver consistent ASF-compliant releases
• Embed transparency and consensus in their processes

Case Study 2: Apache Wave - Stalled Podling and Retirement

Entry

• Originated at Google as “Wave in a Box.”
• Entered the Apache Incubator in December 2010.

During Incubation

• Community stagnation: Most contributions came from a small set of ex-Google engineers; few new committers joined.
• Releases: Rare and inconsistent; struggled with ASF release mechanics.
• Culture: Much discussion remained off-list; limited adoption of ASF decision-making norms.
• Mentorship: Repeated IPMC concerns about diversity and sustainability.

Retirement

After seven years without progress, Wave was retired in April 2018.
Current Status: Retired Podling

Takeaway

Even a high-profile donation can fail if a podling does not establish a sustainable, diverse community and adopt an ASF-style transparency approach.

Case Study 3: Apache Superset - Struggles and Recovery

Entry

• Created at Airbnb.
• Entered the Apache Incubator in July 2017.

During Incubation

• Early struggles: Development focused on GitHub; mailing-list culture was weak. The first incubating release did not arrive until 2019.
• Mentorship: Mentors and IPMC pressed the podling to rotate release managers and move decisions on-list.
• Recovery: By 2020, the community had diversified beyond Airbnb, improved mailing-list engagement, and established a monthly release cadence.

Graduation

Graduated in November 2020 (TLP announcement early 2021).
Current Status: TLP

Takeaway

Superset shows that a podling can recover from early challenges if mentors persist, the community adopts ASF release discipline, and decision-making shifts to transparent forums.

Case Study 4: Apache Flink - Rapid Graduation

Entry

• Originated as the Stratosphere research project at TU Berlin.
• Entered the Apache Incubator in April 2014.

During Incubation

• Releases: The first incubating release was shipped the same year.
• Community: Already had strong multi-organization contributors and a clear academic-to-industry transition.
• Culture: Adopted mailing-list governance and voting practices immediately.

Graduation

Graduated as a Top-Level Project in December 2014.
Current Status: TLP

Takeaway

Flink demonstrates that incubation can be very short when a project enters with an existing, well-organized, and diverse community that quickly adopts ASF practices.

Case Study 5: Apache Iceberg - Building a New Standard

Entry

• Created at Netflix and donated to the ASF.
• Entered the Apache Incubator in November 2018.

During Incubation

• Community growth: Quickly attracted contributors from Apple, Alibaba, Adobe, LinkedIn, and independent developers.
• Releases: Produced multiple incubating releases beginning in 2019.
• Ecosystem impact: Broad adoption via integration into Spark, Flink, and Trino attracted new contributors.
• Mentorship: Mentors provided guidance on branding and ensuring vendor-neutral governance.

Graduation

Graduated as a Top-Level Project in May 2020.
Current Status: TLP

Takeaway

Iceberg shows that projects with strong corporate origins can graduate healthily when they diversify, deliver ASF-compliant releases, and maintain vendor neutrality.

Case Study 6: Apache Blur - Long Stagnation, Retired

Entry

• Search engine built on Lucene.
• Entered the Apache Incubator in July 2012.

During Incubation

• Mailing-list activity and releases remained minimal.
• The community never grew beyond a handful of individuals.
• Repeated IPMC concerns were raised about lack of progress.

Retirement

Retired from incubation in April 2017.
Current Status: Retired Podling

Takeaway

Blur shows that inactivity and failure to build a community often lead to retirement, no matter how promising the technology.

Case Study 7: Apache MXNet - Vendor Dominance and Retirement

Entry

• Originally developed at the University of Washington, later heavily backed by AWS.
• Entered the Apache Incubator in September 2017.

During Incubation

• Releases: Produced multiple incubating releases in the 1.x series under the ASF process.
• Community growth: Activity was concentrated among contributors employed by a single major corporate participant.
• ASF culture challenges: Development relied heavily on GitHub workflows, with relatively low mailing-list participation.
• Mentorship: IPMC discussions raised ongoing concerns about community diversity and long-term sustainability.

Graduation

Graduated as a Top-Level Project in September 2022.

After Graduation

• Contributor diversity did not improve significantly.
• As activity declined, the ASF Board approved retirement.
• Retired to the Apache Attic in September 2023.

Current Status: Retired TLP

Takeaway

MXNet illustrates the risks of concentrated corporate influence. Even with a strong technical foundation and corporate backing, a project cannot thrive at Apache without multi-organization contributor diversity, mailing-list-based consensus, and shared governance.

Case Study 8: Apache Airflow - From Company Project to Global Community

Entry

• Created at Airbnb in 2014 as a workflow orchestration system.
• Entered the Apache Incubator in March 2016.

During Incubation

• Releases: Began producing incubating releases in 2016; later established a steady cadence.
• Community growth: Quickly broadened participation beyond Airbnb - contributors from Google, Polidea, and independents became committers.
• Governance: Shifted all decision-making onto dev@ mailing lists.
• Ecosystem impact: Widely adopted in data-engineering ecosystems; integrations accelerated growth.
• Mentorship: Mentors emphasized neutrality and rotating release managers.

Graduation

Graduated as a Top-Level Project in January 2019.
Current Status: TLP

Takeaway

Airflow illustrates how a project that begins at a single company can evolve into a vibrant, multi-organization community. Keys to success:

• Rotating release managers
• Strong mailing-list culture
• Early outreach beyond the founding organization

Case Study 9: Apache Zipkin - Withdrawing from the Incubator

Entry

• Distributed tracing system, originally created at Twitter in 2012.
• Entered the Apache Incubator in July 2018.

During Incubation

• Community context: An established ecosystem already existed under the OpenZipkin GitHub organization.
• Challenges: Development and issue tracking remained primarily in the external GitHub organization; ASF mailing lists were under-utilized.
• Releases: ASF releases were produced.
• Mentorship: Concerns were raised about lack of ASF alignment (releases, infrastructure, governance).

Outcome

After approximately three years without progress toward graduation, the community voted to withdraw from the Incubator in June 2021.
Development continues outside Apache.
Current Status: Withdrawn Podling

Takeaway

Zipkin illustrates that incubation can be a trial of fit. Sometimes even a popular project determines that continuing outside the ASF model better serves its goals; in such cases, withdrawal can be an appropriate outcome.

Case Study 10: Apache Trafodion - Graduation Followed by Retirement

Entry

• Donated by HP as a transactional SQL-on-HBase database.
• Entered the Apache Incubator in May 2015.

During Incubation

• Releases: Produced multiple incubating releases (2016–2017).
• Governance: Demonstrated readiness and adoption of ASF processes.
• Mentorship: Mentors encouraged the community to continue improving contributor diversity and visibility on mailing lists.

Graduation

Graduated as a Top-Level Project in May 2017.

After Graduation

• Contributor diversity remained limited; community activity slowed.
• The project was retired to the Attic in March 2021.

Current Status: Retired TLP

Takeaway

Graduation is not sufficient for survival. Without long-term diversity and a steady inflow of contributors, even graduated TLPs may be retired.

Case Study 11: Apache NetBeans - Migrating a Mature Codebase

Entry

• Long-standing IDE originally developed by Sun Microsystems and later maintained by Oracle.
• Donated by Oracle and entered the Apache Incubator in October 2016.

During Incubation

• IP clearance and migration: Millions of lines of code required extensive clearance and migration.
• Releases: The first Apache release was NetBeans 9.0 in 2018, nearly two years after entry.
• Community growth: Built governance under ASF norms; contributors beyond Oracle became more active.
• Mentorship: Guided the project in adapting workflows to ASF mailing lists and release practices.

Graduation

Graduated as a Top-Level Project in April 2019.
Current Status: TLP

Takeaway

NetBeans shows that incubation isn’t just for new projects - even mature, widely used codebases must:

• Complete IP clearance at ASF standards
• Transition governance to ASF lists
• Demonstrate the ability to make ASF-compliant releases

Case Study 12: Apache ECharts - Global and Multilingual Community

Entry

• JavaScript-based data-visualization library created at Baidu.
• Entered the Apache Incubator in August 2018.

During Incubation

• Community growth: Expanded beyond Baidu to include contributors from universities, companies, and the broader open-source community.
• Language and inclusivity: Contributors primarily used Chinese early on, and official decision-making moved to English on ASF lists to support transparency and wider participation.
• Releases: Produced multiple incubating releases under ASF processes.
• Governance: Mailing-list culture matured, shifting from external/private channels to ASF infrastructure.
• Outreach: Documentation expanded in both English and Chinese.

Graduation

Graduated as a Top-Level Project in July 2021.
Current Status: TLP

Takeaway

ECharts demonstrates that ASF incubation is effective for globally popular projects, even when most contributors start in a non-English environment. Inclusivity and multilingual practices are vital for sustaining global diversity.

Case Study 13: Apache Hadoop - Ecosystem Catalyst

Entry

• Originated as part of Yahoo!’s search infrastructure and donated to the ASF.
• Entered the Apache Incubator in January 2006.

During Incubation

• Community: Rapidly grew with participation from Yahoo!, Cloudera, Facebook, and others.
• Releases: Produced multiple incubating releases leading to widespread adoption.
• Mentorship: IPMC members provided early guidance on structuring subprojects and clarifying governance boundaries.

Graduation

Graduated as a Top-Level Project in January 2008.
Current Status: TLP

Takeaway

Hadoop showed that transparent, collaborative governance at ASF scale can catalyze entire technology ecosystems.

Case Study 14: Apache Cassandra - From Corporate to Community

Entry

• Initially developed at Facebook and open-sourced in 2008.
• Entered the Apache Incubator in March 2009.

During Incubation

• Releases: Quickly shipped its first incubating release.
• Community growth: Attracted developers from Digg, Rackspace, and independent contributors.
• Mentorship: Mentors helped the community adopt mailing-list–based discussions and ASF voting practices.

Graduation

Graduated as a Top-Level Project in February 2010.
Current Status: TLP

Takeaway

Cassandra’s evolution shows that early establishment of transparent, multi-organization governance is essential to maintaining community neutrality and resilience against vendor concentration over time.

Case Study 15: Apache Kafka - Vendor-Neutral Governance at Scale

Entry

• Created at LinkedIn as a distributed streaming platform.
• Entered the Apache Incubator in July 2011.

During Incubation

• Releases: Delivered its first incubating release in 2011 and established a regular cadence.
• Community growth: Contributors expanded to include Confluent, Netflix, Yahoo!, and independents.
• Mentorship: Mentors advised on standardizing release voting and communication practices that later influenced other ASF projects.

Graduation

Graduated as a Top-Level Project in October 2012.
Current Status: TLP

Takeaway

Kafka is often cited as a strong multi-company community; as its commercial ecosystem expanded, the community continued to discuss how to preserve vendor-neutral governance.

Case Study 16: Apache Pulsar - Modern Distributed Messaging with Transparent Governance

Entry

• Originally developed at Yahoo! for cloud messaging.
• Entered the Apache Incubator in June 2017.

During Incubation

• Releases: First incubating release in 2017 with consistent cadence thereafter.
• Community growth: Expanded to Splunk, StreamNative, Tencent, and independent contributors.
• Mentorship: Mentors supported the community in balancing corporate and independent participation within ASF governance norms.

Graduation

Graduated as a Top-Level Project in September 2018.
Current Status: TLP

Takeaway

Pulsar’s steady releases and strong mentor collaboration show how consistent transparency accelerates maturity and trust.

Case Study 17: Apache APISIX - Cloud-Native Global Growth

Entry

• Developed by Shenzhen-based API7.ai as an open-source API gateway.
• Entered the Apache Incubator in October 2019.

During Incubation

• Releases: Produced frequent releases, emphasizing security and stability.
• Community growth: Attracted contributors from China, Europe, and the US within the first year.
• Governance: Moved from company-driven decisions to transparent voting on dev@.
• Mentorship: Mentors stressed language accessibility and neutral branding.

Graduation

Graduated as a Top-Level Project in July 2020.
Current Status: TLP

Takeaway

APISIX highlights ASF’s ability to host fast-moving, globally distributed, cloud-native projects that balance corporate and independent participation.

Case Study 18: Apache IoTDB - Academic Research to Industrial Adoption

Entry

• Created at Tsinghua University as part of the TSFile research project.
• Entered the Apache Incubator in November 2018.

During Incubation

• Releases: First incubating release in 2019 and multiple updates through 2020.
• Community growth: Expanded to contributors from Huawei, Alibaba, and Europe.
• Mentorship: Encouraged improved documentation and governance in English.

Graduation

Graduated as a Top-Level Project in September 2020.
Current Status: TLP

Takeaway

IoTDB shows that academic research can evolve into sustainable open-source communities when they adopt transparent governance and regular release practice.

Case Study 19: Apache Spark - From Research to Broad Adoption

Entry

• Originated at UC Berkeley’s AMPLab and open-sourced in 2010.
• Entered the Apache Incubator in June 2013.

During Incubation

• Community growth: From the outset, Spark attracted contributors from Databricks, Cloudera, Intel, and independent researchers.
• Releases: Produced multiple incubating releases across 2013 and 2014 with frequent testing cycles.
• Mentorship: Mentors emphasized clear review processes and consistent release voting.
• Adoption: Widely integrated into Hadoop ecosystems, driving community expansion.

Graduation

Graduated as a Top-Level Project in February 2014.
Current Status: TLP

Takeaway

Spark demonstrates that academic innovation can scale to global enterprise adoption when guided by transparent governance and frequent ASF-compliant releases.

Case Study 20: Apache Storm - Early Stream-Processing Success

Entry

• Created at BackType and later acquired by Twitter.
• Entered the Apache Incubator in September 2013.

During Incubation

• Releases: Delivered its first ASF-compliant release in 2014 after license audits.
• Community growth: Expanded beyond Twitter to contributors at Yahoo!, Spotify, and independent developers.
• Mentorship: Mentors encouraged simplifying voting and release sign-offs for a small team.
• Adoption: Recognized as a pioneer of real-time stream processing in open source.

Graduation

Graduated as a Top-Level Project in September 2014.
Current Status: TLP

Takeaway

Storm was early in real-time stream processing at Apache, showing that small teams can succeed when they fully embrace ASF release discipline and transparent development. Now in a stable maintenance phase, it continues to demonstrate the value of long-term stewardship within the ASF model.

Case Study 21: Apache Mesos - Cluster Management at Scale

Entry

• Originated as a UC Berkeley research project.
• Entered the Apache Incubator in December 2010.

During Incubation

• Community growth: Became a core component in many large-scale infrastructure deployments at Twitter and Airbnb.
• Releases: Frequent incubating releases validated its production stability.
• Mentorship: Mentors guided early multi-org collaboration and governance rules.

Graduation

Graduated as a Top-Level Project in July 2013.

After Graduation

• Activity declined as industry focus shifted to Kubernetes.
• The project was retired to the Apache Attic in October 2021.

Current Status: Retired TLP

Takeaway

Mesos was once core infrastructure at scale; its retirement shows that sustaining relevance requires continuous community renewal and adaptation to industry change.

Case Study 22: Apache Druid - High-Performance OLAP at ASF Scale

Entry

• Originally created at Metamarkets for real-time analytics.
• Entered the Apache Incubator in February 2018.

During Incubation

• Releases: Produced multiple incubating releases through 2019.
• Community growth: Expanded to contributors from Imply, Alibaba, and other data companies.
• Mentorship: Mentors focused on ensuring ASF-compliant release processes and governance.
• Adoption: Quickly adopted in cloud and analytics platforms.

Graduation

Graduated as a Top-Level Project in December 2019.
Current Status: TLP

Takeaway

Druid demonstrated that ASF incubation can support enterprise-focused, cloud-native analytics software while maintaining neutral governance.

Case Study 23: Apache Parquet - Shared Foundation for Analytics

Entry

• Originally developed by Twitter and Cloudera for columnar data storage.
• Entered the Apache Incubator in April 2014.

During Incubation

• Releases: Completed multiple incubating releases to stabilize the format specification.
• Community growth: Adopted by projects like Impala, Drill, and Spark.
• Mentorship: Focused on maintaining neutrality for a shared format used across ASF ecosystem projects.

Graduation

Graduated as a Top-Level Project in April 2015.
Current Status: TLP

Takeaway

Parquet illustrates how incubation can steward a neutral, shared standard that supports the broader ASF analytics ecosystem.

Case Study 24: Apache Arrow - Cross-Language Interoperability

Entry

• Founded as a collaboration among multiple ASF projects to create a shared in-memory columnar format.
• Entered the Apache Incubator in February 2016.

During Incubation

• Community growth: Included contributors from Dremio, Ursa Computing, and independent developers.
• Releases: Delivered regular incubating releases and cross-project integration tests.
• Mentorship: Emphasized cross-language standardization and API consistency governed through ASF lists.

Graduation

Graduated as a Top-Level Project in February 2018.
Current Status: TLP

Takeaway

Arrow shows that ASF governance can successfully coordinate long-term, cross-language data standards serving a broad ecosystem.

Case Study 25: Apache NiFi - From Government Research to Open Collaboration

Entry

• Originated within the U.S. National Security Agency (NSA) as “NiagaraFiles.”
• Donated to the ASF through a U.S. government release process.
• Entered the Apache Incubator in November 2014.

During Incubation

• Releases: Produced multiple incubating releases by 2015.
• Community growth: Expanded beyond government contributors to independent and industry participants.
• Mentorship: Mentors ensured transparent review of export control and licensing concerns.
• Governance: Established dev@ communication norms for handling sensitive topics openly.

Graduation

Graduated as a Top-Level Project in July 2015.
Current Status: TLP

Takeaway

NiFi proved that government-developed software can thrive under ASF transparency and community-driven collaboration.

Case Study 26: Apache Kylin - Early China-Initiated ASF TLP

Entry

• Developed at eBay China for OLAP on Hadoop.
• Entered the Apache Incubator in November 2014.

During Incubation

• Releases: Produced multiple incubating releases through 2015.
• Community growth: Added committers from Kyligence, Huawei, and independent developers.
• Mentorship: Focused on cross-time-zone communication and English-language transparency.
• Adoption: Widely used in Asia and globally as an OLAP engine.

Graduation

Graduated as a Top-Level Project in December 2015.
Current Status: TLP

Takeaway

Kylin was an important milestone in ASF globalization, demonstrating that a China-based team can succeed by embracing The Apache Way of transparency and inclusive governance.