What software license should I choose for enterprise open source? " In the article, the significance of different agreements for enterprise open source has been discussed from the perspective of software licenses.
However, that article mainly started from the content of the license, that is, the agreement was first given, the details of the agreement were analyzed, and then it was judged whether it conformed to the situation of the enterprise and the software planned to be released. This article starts from the different appeals and risks of enterprise open source, combined with the existing enterprise organization practice, to re-discuss how to choose the software protocol that the first step of source code disclosure should face.
completely open source
Completely open source, that is, the situation in which the enterprise's self-developed software is released with a software protocol that conforms to the definition of open source [1 ] .
Before discussing what kind of enterprise will adopt a completely open source strategy for what kind of software, we first discuss what risks a completely open source strategy will bring to the enterprise. Software agreements that meet the definition of open source include both permissive software agreements and copyleft-style agreements, and no distinction is made here because enterprises face the same problems.
Of course, for the situation where the enterprise completely owns the property rights, the enterprise can change the software agreement at any time to avoid the risks mentioned below. So the risks mentioned here mainly involve donating self-developed software to third parties such as open source foundations, or companies trying to adhere to a completely open source strategy.
Risk: Business competition from competitors
I am in " Hackers and Customers: Can Open Source Software Be Commercialized?" " As discussed in the article, users of open source software include two types: hackers and customers.
For the founding team of completely open source code, since hacker users have the ability to use the software, or even modify the software to meet their own needs, it can be considered that the development costs have been paid, and instead of making money from such users, seek cooperation Joint construction is reasonable. At present, the logic of most "open source commercialization" enterprises is also based on customers who are unable to serve or unwilling to undertake development and maintenance tasks.
This can be seen from the fact that many practices of the emerging Elastic License 2.0 [2] (ELv2) and Business Source License 1.1 [3] (BSLv1) allow free use without competition of similar commercial services. The above two commercial software agreements with open source code want to avoid commercial competition from fellow hackers.
For example, there are countless companies in the industry that offer managed Apache Kafka services. In addition to Confluent [4] , which has gathered a considerable number of core developers , SaaS companies such as Upstash [5 ] and Aiven [6] all provide Kafka hosting services, not to mention the large and small platform cloud vendors. Kafka services, such as AWS's MSK [7 ] service, Alibaba Cloud's Kafka service [8 ] and Red Hat's OpenShift Streams for Apache Kafka [9] service, etc.
Such competition has forced Confluent to seek new growth points beyond simply providing Kafka services, such as providing multi-tenant functions only in the commercial version, and many failed attempts at stream computing integration. Of course, Kafka's founding work and donations are all done under the leadership of LinkedIn Corporation. In this way, although Confluent cannot imitate MongoDB and change the agreement when competitors directly provide commercial services, but it is not too unbalanced.
ElasticSearch and MongoDB were originally released under Apache License 2.0 [10] (ALv2) and GNU Affero General Public License 3.0 [11] (AGPLv3), but under the competition of commercial companies such as AWS with low technology costs and high platform advantages, neither The new increase focuses on alleviating the crisis, but it is unable to resist the imminent dredging plan, so the change agreement directly prohibits the use of source code for commercial competition.
Risk: Team splits and competes
The case of Confluent implies another risk of complete open source, that is, when the software can be freely interpreted and commercialized, the developers in the enterprise may split off to conduct independent business operations and participate in competition.
The founding team of Confluent comes from LinkedIn. LinkedIn itself does not provide cloud services, and this kind of competition is not yet obvious. Other companies in the LinkedIn situation can even provide early investment in exchange for lower-cost support for new teams than direct hires before.
The story of Apache Hadoop demonstrates the risks of multiple evenly-matched companies, while doing simple encapsulation of commercial models based on non-proprietary open source software.
With Hadoop's incubator Yahoo not in the market, Cloudera, Hortonworks, and MapR competed for Hadoop's privatized deployment and operation and maintenance services. The final outcome was that Hortonworks and MapR were acquired, and Cloudera was privatized. delisted. However, after delisting, Cloudera still faces fierce competition from providers on the market who have Apache Ambari packages or self-developed big data suites.
The story of Apache Flink illustrates the risk of fragmentation when some teams seek independence when one company dominates the community.
Initially, Flink was developed at a German university and donated to the Apache Software Foundation. The founding members then established Ververica, which was acquired by Alibaba in 2019. Alibaba, which has completed the acquisition, has basically formed the dominance of the Flink community, and will not need to worry about competition from other Flink-based commercial services for a period of time to come. However, in the past one or two years, the core team left several times to join other companies that provide commercial services based on Flink, or set up a separate commercial company to provide Flink services. Among them, Immerok [12] , which took the latter path , was acquired by Confluent not long ago, which suddenly made Alibaba's stream computing business strongly blocked by Confluent stream storage + stream computing.
Behind these dramatic business operations, Flink itself is a completely open-source software that allows anyone and organization to freely interpret and use it, including commercial use and providing commercial competition, which is a basic support for the compliance level.
In the examples mentioned above, even if there are cases where enterprises have completed the dominant discourse power of the community, they are all cases where enterprises do not have independent property rights to open source software. Although the following example is also a project of the Apache Software Foundation, the method used in it can be applied to the software with its own property rights.
This case is the Apache Doris project.
Apache Doris was developed by Baidu's R&D team and donated to the Apache Software Foundation. With its design of data analysis capabilities that meet the needs of domestic enterprises, Doris has gradually developed a considerable user group in the first ten years, a considerable part of which is still a potential customer for purchasing commercial services. Baidu released its Palo [13] data analysis product based on Doris, but soon after, some core developers left and forked Doris to release a similar competitive product that was later renamed StarRocks [14] . Within a few years, another R&D team left and established SelectDB [15] to provide similar data analysis services based on Doris.
Just looking at the competition around Doris, it seems that it is no different from the competition of several Hadoop companies. But StarRocks is not a typical version around the upstream, but a brand new version that is hard-branched from a certain version. The current relationship with Doris can be regarded as "the road is open, each side goes to the other side". The competition brought about by such a hard branch method cannot be avoided by modifying the agreement even if the upstream software is completely self-owned.
In order to show that the case of Doris is not an isolated case, here are a few more similar business-related hard branch cases:
• Trino is a hard fork based on Facebook Presto, followed by Trino team established Starburst [16] The company proposes the concept of Data Mesh as a commercialization direction.
•Lightbend[17] After the company switched Akka to use by companies with annual revenues of more than $10 million and required a commercial agreement, neutral members of the Akka community forked Akka and continued to run it under the name Pekko [18] in the Apache Software Foundation's incubator.
• AWS launched the OpenSearch [19] project after Elastic Company switched ElasticSearch to ELv2 and provided similar commercial services based on it.
In the next section, I describe how companies adopting a fully open source model can address these risks without changing the software co-model.
References
[1] Open Source Definition: https://opensource.org/definition/[2] Elastic License 2.0: https://www.elastic.co/licensing/elastic-license[3] Business Source License 1.1: https://mariadb.com/bsl11/[4] Confluent: https ://www.confluent.io/[5] Upstash: https://docs.upstash.com/kafka[6] Aiven: https://aiven.io/kafka[7] MSK: https://aws.amazon.com/msk/[8] Alibaba Cloud Kafka service: https://www.alibabacloud.com/product/kafka[9] OpenShift Streams for Apache Kafka: https://www.redhat.com/en/technologies/cloud-computing/openshift/openshift-streams-for-apache- Kafka[10] Apache License 2.0: https://www.apache.org/licenses/LICENSE-2.0[11] GNU Affero General Public License 3.0: https://www.gnu.org/licenses/agpl-3.0.en.html[12] Immerok: https://www.immerok.io/[13] Palo: https://intl.cloud.baidu.com/product/palo.html[14] StarRocks: https://www.starrocks.io/[15] SelectDB: https://en.selectdb.com/[16] Starburst: https://www.starburst.io/[17] Lightbend: https://www.lightbend.com/[18] Pekko: https://pekko.apache.org/[19] OpenSearch: https://opensearch.org/
Author丨tisonkun
Editor丨Li Jiaxin
Design丨Zhao Yuyue
Related Reading | Related Reading
software freedom to free software
Hackers and Customers Can open source software be commercialized?
Introduction to Kaiyuanshe
Founded in 2014, Kaiyuan Society is composed of individual members who voluntarily contribute to the cause of open source. It is formed according to the principle of "contribution, consensus, and co-governance". It has always maintained the characteristics of vendor neutrality, public welfare, and non-profit. International integration, community development, project incubation" is an open source community federation with the mission. Kaiyuanshe actively cooperates closely with communities, enterprises and government-related units that support open source. With the vision of "Based in China and Contributing to the World", it aims to create a healthy and sustainable open source ecosystem and promote China's open source community to become an active force in the global open source system. Participation and Contributors.
In 2017, Kaiyuanshe was transformed into an organization composed entirely of individual members, operating with reference to the governance model of top international open source foundations such as ASF. In the past nine years, it has connected tens of thousands of open source people, gathered thousands of community members and volunteers, hundreds of lecturers at home and abroad, and cooperated with hundreds of sponsors, media, and community partners.
、