Lars Kurth 
Community Manger, Xen Project 
Chairman, Xen Project Advisory Board 
Director, Open Source Business Office, Citrix lars_kurth
Was a contributor to various projects 
Worked in parallel computing, tools, 
mobile and now virtualization 
Community guy at Symbian Foundation 
Learned how NOT to do stuff 
Community guy for the Xen Project 
Working for Citrix 
Member of OSS Business Office 
Accountable to Xen Project Advisory Board 
Chairman of Xen Project Advisory Board
Source: The 2014 Future of Open Source Survey Result
The # of Projects is growing rapidly 
2007: 0.2M projects 
Today: 1.0M projects, 
100Billion LOC, 
10M contributors 
2015: 1.8M projects 
John Morgan @ Flickr 
Simon & His Camera @ Flickr 
Kumar Appaiah @ Flickr
50% of all enterprises adopt OSS 
software 
Julian Manson @ Flickr
30% of companies make it easy for 
employees to contribute to projects 
Influencing a project’s direction is one 
of the main reason for contributing 
Nick @ Flickr 
toffehoff @ Flickr
Open Source Flywheel
[1] bit.do/optionvalue
2001: Open Sourced by IBM 
2001 – 2003: Growth from 8 to 80 
consortia members 
2006: Callisto 
10 projects, 260 committers 
2004: Eclipse Foundation 
2011: 10th Birthday 
More projects/products/users, 
improved process, improved 
option value/modularity, … 
273 projects, 1057 committers, 
50+ MLOC, 174 members 
(see bit.do/Eclipse-10) 
…
Tragedy of the Commons 
(sort of) 
Moyan Brenn @ Flickr
snoopsmouse @ Flickr
Catastrophic is the right word 
[for Heartbleed]. On the scale 
of 1 to 10, this is an 11. 
Bruce Schneier 
Internationally renowned security Technologist 
@Bruce_Schneier
Source: Ohloh.net 
Growing Codebase 
Static and small contributor base 
1 person maintaining 100 KLoC = 
Underinvestment 
Extremely large user base 
Critical infrastructure component 
Thus impact of Heartbleed is huge
snoopsmouse @ Flickr
Features 
How many users you have 
How many vendors back you 
How you are seen in the press 
…
can help you succeed
Neutrality / Perception 
Support Infrastructure 
Expertise / Mentoring 
Vendor Network 
… 
BUT: You still need to do 
all the right things
Create a balanced Flywheel 
OSS Foundations can help 
Master different skills 
Vinovyn @ Flickr
Follow Industry News 
Follow Project News 
Trial Software 
Adopt Software 
Engage with Users 
Engage with Industry 
Evangelize 
Customize 
Contribute 
Lead 
Problem 
Activities 
Metrics
Problem 
Community is expanding fast 
Companies can’t find staff 
Potential of slowing growth 
Potential of loosing new vendors 
Poaching 
Friction 
Tim J Keegan @ Flickr
Training: 
Expensive to create 
Does not normally scale 
(unless your project is huge) 
bit.do/grow-devs 
You don’t need training… 
Good and active dev wiki 
In code API docs 
Recorded developer talks 
Extensive commit messages 
Design discussions on the list 
(linked to in commit message) 
Willingness to help newcomers 
(mentors) 
...
Orgs need to apply 
(we did not always get in) 
Students only, summer only 
Orgs and companies can sponsor 
slots 
Interns, summer and winter 
Good mentors = word of mouth 
Committed mentors (who can spend a few hours a week) 
Suitable projects 
Not so much about the code = it’s an investment in people 
Metrics 
Approx. 40% still engaged after 3 years 
Repeat mentors, more people willing to mentor
Problem 
30% Community Growth p.a. 
Contributors competing for review 
time from stretched maintainer / 
reviewer base 
Average review time up from 28 to 
32 days in 6 months
Short term fixes 
(documentation/guidance : set expectations) 
(better co-ordination for amongst reviewers) 
(focus more on designs to save time) 
Grow reviewer base by identifying capable candidates in the community 
(support + encourage + reward : bit.do/let-them-be-your-heroes) 
Mentor candidates to get them effective more quickly than normal 
(copy the GSoC / OPW pattern) 
Get backing from vendors to ensure candidates stay engaged in community 
(if vendor employee) 
(existing community members don’t want to waste their time on someone 
who will dissappear)
Problem 
Activities 
Metrics
Activities 
Metrics 
Problem
Industry 
Media 
Media Coverage 
Social Media Impact
Users 
Industry 
Media 
User & Web Metrics 
Features / Releases
Community Metrics 
bit.do/oss-metrics 
bit.do/oss-commercial-dashboards 
Contributors 
Users 
Industry 
Media
Are you helping 
your employer 
by contributing to 
the project? 
Contributors 
Employer 
Users 
Industry 
Media
Are you spending 
your funds wisely? 
bit.do/community-plan 
Contributors 
Employer 
Board 
Users 
Industry 
Media
Open Source Strategy / 
Community Plan 
bit.do/sample_community_plan
bit.do/xenproject-sample-report-1 
bit.do/xenproject-sample-report-2
Open Source Flywheel 
revisited 
The boundary between 
Business and Communities
Create a balanced Flywheel 
OSS Foundations can help 
Master many different skills 
Vinovyn @ Flickr 
Get Companies to help you
Undefined Responsibilities 
cause problems 
Moyan Brenn @ Flickr
Xen Project Advisory Board trying 
to push a preferred test harness 
over community solution 
Paralysis: no new test code 
written 
Delay of roll-out of independently 
hosted Test Farm 
Risk of Test Farm not being 
adopted 
• Working group jointly led by community and Advisory Board 
• Group resolved the issue
Vendor trying to use private 
channels to Citrix Xen Project 
maintainers to get an edge 
Committers needed the vendor 
help to progress their goals 
Vendor trying to get more and 
more 
Potential of lack of trust in our 
Xen Project maintainers 
• A rather difficult conversation 
• Vendor starting to follow community practices and additionally donating 
(non-developer) resources to the project
Tedious and Business 
Unfriendly Rules 
Moyan Brenn @ Flickr
Vendors wants to promote 
project at events (swag, booth, 
collateral, …) 
Tedious approval process 
for every single instance 
• PMC approval 
• VP of Trademark approval 
(bottleneck) 
Frustrated vendors 
Frustrated community 
• Simplify process for common situations 
• Proposal at bit.do/PMC-TM-management
Companies Community 
Advisory Board WGs Project Governance 
• PR / AR / Marketing / 
Messaging 
• Membership Rules / 
Trademarks / Legal 
• Provide funds to solve 
“Common Good” problems 
• Referee of last resort 
• Principles (aka Values) 
• Roles 
• Decision Making 
• Project Lifecycle 
• Community Initiatives, Best 
Practices, … 
PR WG 
Test WG 
TCT
Clear Rules and Responsibilities 
Remove tension 
Create trust 
Works best when aligned with Flywheel 
Business friendly Rules 
Simple and Easy 
Enables businesses to help the community 
Effect: Community sees value in company 
participation 
Vinovyn @ Flickr
Vinovyn @ Flickr
Maria Ly @ Flickr
Tchami @ Flickr
Please rate the talk 
www.slideshare.net/xen_com_mgr/

LCEU14: How to run a Collaborative Project - Lars Kurth

  • 1.
    Lars Kurth CommunityManger, Xen Project Chairman, Xen Project Advisory Board Director, Open Source Business Office, Citrix lars_kurth
  • 2.
    Was a contributorto various projects Worked in parallel computing, tools, mobile and now virtualization Community guy at Symbian Foundation Learned how NOT to do stuff Community guy for the Xen Project Working for Citrix Member of OSS Business Office Accountable to Xen Project Advisory Board Chairman of Xen Project Advisory Board
  • 3.
    Source: The 2014Future of Open Source Survey Result
  • 4.
    The # ofProjects is growing rapidly 2007: 0.2M projects Today: 1.0M projects, 100Billion LOC, 10M contributors 2015: 1.8M projects John Morgan @ Flickr Simon & His Camera @ Flickr Kumar Appaiah @ Flickr
  • 5.
    50% of allenterprises adopt OSS software Julian Manson @ Flickr
  • 6.
    30% of companiesmake it easy for employees to contribute to projects Influencing a project’s direction is one of the main reason for contributing Nick @ Flickr toffehoff @ Flickr
  • 7.
  • 9.
  • 14.
    2001: Open Sourcedby IBM 2001 – 2003: Growth from 8 to 80 consortia members 2006: Callisto 10 projects, 260 committers 2004: Eclipse Foundation 2011: 10th Birthday More projects/products/users, improved process, improved option value/modularity, … 273 projects, 1057 committers, 50+ MLOC, 174 members (see bit.do/Eclipse-10) …
  • 15.
    Tragedy of theCommons (sort of) Moyan Brenn @ Flickr
  • 16.
  • 17.
    Catastrophic is theright word [for Heartbleed]. On the scale of 1 to 10, this is an 11. Bruce Schneier Internationally renowned security Technologist @Bruce_Schneier
  • 19.
    Source: Ohloh.net GrowingCodebase Static and small contributor base 1 person maintaining 100 KLoC = Underinvestment Extremely large user base Critical infrastructure component Thus impact of Heartbleed is huge
  • 22.
  • 23.
    Features How manyusers you have How many vendors back you How you are seen in the press …
  • 24.
    can help yousucceed
  • 25.
    Neutrality / Perception Support Infrastructure Expertise / Mentoring Vendor Network … BUT: You still need to do all the right things
  • 27.
    Create a balancedFlywheel OSS Foundations can help Master different skills Vinovyn @ Flickr
  • 28.
    Follow Industry News Follow Project News Trial Software Adopt Software Engage with Users Engage with Industry Evangelize Customize Contribute Lead Problem Activities Metrics
  • 29.
    Problem Community isexpanding fast Companies can’t find staff Potential of slowing growth Potential of loosing new vendors Poaching Friction Tim J Keegan @ Flickr
  • 30.
    Training: Expensive tocreate Does not normally scale (unless your project is huge) bit.do/grow-devs You don’t need training… Good and active dev wiki In code API docs Recorded developer talks Extensive commit messages Design discussions on the list (linked to in commit message) Willingness to help newcomers (mentors) ...
  • 31.
    Orgs need toapply (we did not always get in) Students only, summer only Orgs and companies can sponsor slots Interns, summer and winter Good mentors = word of mouth Committed mentors (who can spend a few hours a week) Suitable projects Not so much about the code = it’s an investment in people Metrics Approx. 40% still engaged after 3 years Repeat mentors, more people willing to mentor
  • 32.
    Problem 30% CommunityGrowth p.a. Contributors competing for review time from stretched maintainer / reviewer base Average review time up from 28 to 32 days in 6 months
  • 33.
    Short term fixes (documentation/guidance : set expectations) (better co-ordination for amongst reviewers) (focus more on designs to save time) Grow reviewer base by identifying capable candidates in the community (support + encourage + reward : bit.do/let-them-be-your-heroes) Mentor candidates to get them effective more quickly than normal (copy the GSoC / OPW pattern) Get backing from vendors to ensure candidates stay engaged in community (if vendor employee) (existing community members don’t want to waste their time on someone who will dissappear)
  • 34.
  • 35.
  • 36.
    Industry Media MediaCoverage Social Media Impact
  • 37.
    Users Industry Media User & Web Metrics Features / Releases
  • 38.
    Community Metrics bit.do/oss-metrics bit.do/oss-commercial-dashboards Contributors Users Industry Media
  • 39.
    Are you helping your employer by contributing to the project? Contributors Employer Users Industry Media
  • 40.
    Are you spending your funds wisely? bit.do/community-plan Contributors Employer Board Users Industry Media
  • 41.
    Open Source Strategy/ Community Plan bit.do/sample_community_plan
  • 42.
  • 43.
    Open Source Flywheel revisited The boundary between Business and Communities
  • 46.
    Create a balancedFlywheel OSS Foundations can help Master many different skills Vinovyn @ Flickr Get Companies to help you
  • 48.
    Undefined Responsibilities causeproblems Moyan Brenn @ Flickr
  • 49.
    Xen Project AdvisoryBoard trying to push a preferred test harness over community solution Paralysis: no new test code written Delay of roll-out of independently hosted Test Farm Risk of Test Farm not being adopted • Working group jointly led by community and Advisory Board • Group resolved the issue
  • 50.
    Vendor trying touse private channels to Citrix Xen Project maintainers to get an edge Committers needed the vendor help to progress their goals Vendor trying to get more and more Potential of lack of trust in our Xen Project maintainers • A rather difficult conversation • Vendor starting to follow community practices and additionally donating (non-developer) resources to the project
  • 52.
    Tedious and Business Unfriendly Rules Moyan Brenn @ Flickr
  • 53.
    Vendors wants topromote project at events (swag, booth, collateral, …) Tedious approval process for every single instance • PMC approval • VP of Trademark approval (bottleneck) Frustrated vendors Frustrated community • Simplify process for common situations • Proposal at bit.do/PMC-TM-management
  • 55.
    Companies Community AdvisoryBoard WGs Project Governance • PR / AR / Marketing / Messaging • Membership Rules / Trademarks / Legal • Provide funds to solve “Common Good” problems • Referee of last resort • Principles (aka Values) • Roles • Decision Making • Project Lifecycle • Community Initiatives, Best Practices, … PR WG Test WG TCT
  • 57.
    Clear Rules andResponsibilities Remove tension Create trust Works best when aligned with Flywheel Business friendly Rules Simple and Easy Enables businesses to help the community Effect: Community sees value in company participation Vinovyn @ Flickr
  • 58.
  • 59.
    Maria Ly @Flickr
  • 61.
  • 64.
    Please rate thetalk www.slideshare.net/xen_com_mgr/

Editor's Notes

  • #2 Lots of concrete examples BUT: when I tried to do this, it would have amounted to dishing the dirt about MANY companies SO I decided to focus on theory and tools instead
  • #3 Love to travel to weird places and grow weird plants
  • #4 HIGHLIGHT SURVEY RESULTS
  • #5 Munich, London, LA
  • #8 TIMING: 7 MINS Inspired by the Bezos Flywheel which “explains how to attract ever more sales in an online store”
  • #9 #1: when you have a working cycle, the outcome is “community growth” So let’s look at this on more detail
  • #10 #1 We all understand the basic elements of the OSS dev model I did want to cover some important but less covered aspects Option Value = ability to implement new feature in the easiest way for the developer Modularity #ALL : the whole cycle can become virtuous ## There are also feedback loops (e.g. Users to Dev Activity) – but I want to keep things simple for now
  • #11 So: how this this actually work?
  • #12 #1: when you have a working cycle, the outcome is “community growth” So let’s look at this on more detail
  • #13 # More biz opportunities => More development activity BECAUSE Existing vendors have more opportunities
  • #14 To accommodate for the whole cycle, the community has to evolve and improve the development process E.g. do new things, such as marketing, test infrastructure, fix problems with reviewing and scale, … Then CLICK: which means betterdev model … Then CLICK
  • #15 I don’t want you to take me by my word – there are actually examples Chose Eclipse, because I was involved from 2006-2009 Birthday stats instead of today’s : didn’t want to count the data myself and there was a published document for 2011 LINUX & OpenStack
  • #16 TIMING: 12 MINS Definition: Individuals, acting independently and rationally according to each one's self-interest, behave contrary to the whole group's long-term best interests by depleting some common resource A lot of us probably dealt with common goods issues such as documentation, testing, … - but I wanted to pick an example of a new class of problems
  • #20 Mention: Core Infrastructure Initiative
  • #21 The reason for these issues is often an imbalanced or broken cycle In the case of the Xen Project, we also had an imbalanced cycle, and I spent the last 4 years at the project fixing this.
  • #22 TIMING: 15 MINS 10 years ago, putting the code out there was usually enough to be successful! Today we have a lot of new challenges
  • #23 Unlimited growth This is not true any more, because we hit resource limits
  • #25 This story is about change : so let’s start with how OSS changed
  • #27 Example: # Apache main strength is the “Apache Way” of building communities # One of the LFs main strength is the mix of commercial focus and marketing capabilities WORRY: it is not yet clear what the rapid increase of extremely well funded projects (examples: AllSeen, OpenDayLight, OPNFV, …) will do to the open source eco-system at large
  • #28 Link back to Jim Zemlin’s keynote about the rise of OSS foundations
  • #29 TIMING: 20 MINS #1 Inspired by Sales Funnels : conversion of leads from media interest to users to developers/companies Effectiveness determined by shape of funnel #2 The shape can be manipulated by activities (e.g. Test Days) and external events (e.g. what the competition does) #3 Don’t have time to go through this in detail, but there is a blog post and presentation with a case study explaining it
  • #30 TIMING: 22 MINS Picked this because this is a problem in my community as well as many Linux Communities This can have a bad effect on your project
  • #32 Word of mouth => 60% of our joint GSoC and OPW participants being women
  • #34 Mentoring does imply that existing maintainers and reviewers spend more time
  • #35 TIMING: 30 MINUTES
  • #42 Connecting the Funnel + Flywheel
  • #44 TIMING: 35 MINUTES
  • #45 Marketing vs. Community & Developer Driven
  • #47 Do you remember that lesson? CLICK … in the areas they do well
  • #48 Unfortunately that is easier said than done --- CLICK: because in practice there is often tension when one tries to do this CLICK: they have different reasons for participating --- BUT: the common interest for both companies and people mostly participate for their own selfish interests In other words: it’s unfair to say the interests of companies vs. those of people are better or worse than the others --- And of course that tension is manageable And some (!) OSS foundations are very good in helping manage that tension
  • #49 TIMING: 38 MINUTES Responsibility in the sense of who should and is allowed to do what
  • #50 The root cause was that we had no proven mechanism for companies and the community to work together
  • #51 NOTE: I also used the conversation to ask for stuff from the vendor (such as donating of HW, …) – and now I am usually setting expectations upfront
  • #52 Common thread to these stories … No bad intentions in any of them Companies and people will always try and push boundaries (in particular when they are fuzzy) to follow their self-interest
  • #53 TIMING: 42 MINUTES
  • #55 TIMING: 40 MINS
  • #56 Cross-over of people, e.g. committers and maintainers on board and WG groups
  • #57 Apache, Commercial Projects, …
  • #58 TIMING: 45 MINUTES
  • #59 TIMING: 34 MINUTES
  • #62 In other words : why not let companies contribute expertise which they have rather than doing it all yourself
  • #63 You must NEVER forget that open source and communities are about PEOPLE and BELONGING
  • #64 BELONGING, FAIRNESS, FUN, …
  • #65 TIMING: 36 MINUTES