AB Meeting/August 2014 Minutes

Attendees

• Lars Kurth (Chair)

• James Bulpin (Citrix)
• Larry Wikelius (Cavium)
• Thomas Molgaard (ARM – verified to be able on behalf of ARM prior to the meeting)
• Don Slutz (Verizon – verified to be able on behalf of Verizon prior to the meeting)
• Konrad Wilk, Daniel Kiper (Oracle)
• Don D Dugger (Intel)
• Matt Wilson (AWS)
• Guido Trotter (Google)
• Sherry Hurwitz (AMD)

Quorum

• Sarah Conway (Linux Foundation)
• Brian Warner (Linux Foundation)

Advisory Board Chairman Ship

I was elected to chair the Advisory Board at last August’s board meeting. I am happy to remain chairman for another year, but we do need to have a formal vote. We can either do this during the meeting. If someone else wanted to step up, we probably need to look at the process – which is somewhat informal right now - and will have to have an email vote after the meeting.

Konrad: nominated Lars Kurth as chair

Vote|Motion:  to vote Lars as chairman for another year = approved

Test Framework – Independent Infrastructure

See detailed proposal and follow-up at

• http://lists.xenproject.org/archives/html/advisory-board/2014-08/msg00014.html
• http://lists.xenproject.org/archives/html/advisory-board/2014-08/msg00015.html

Questions:

• Sherry : what is the proportion of AMD vs. Intel machines
• Lars did not have a clear answer as the detailed machine list was not sent to him
• Don Slutz sent detailed list and it appears that the list is Intel heavy and needs to be adjusted
• Lars proposes to sign off the budget, but that the board and test framework WG proposes any changes to the detailed machine specs. The goal is to balance variety between AMD and Intel

 Action Don & Lars: Send detailed machine list to the list ✓
 Action Lars: Work with test framework WG and board to finalise HW list. Aim to stay within the approved budget

• Larry wanted to ensure that ARM machines can be added later
• Don: we will need to add more cabinet space later on to do this, but we do have the capability to expand the setup as needed in future
• Lars: we have sufficient budget for more space and machines this year and next year due to significant underspend for test framework related services and hardware. So there should be no issue in expanding in due course
• Larry: how quickly can we move forward once the budget is approved
• Don: we can be active in under a month
• Konrad: What is the process moving forward
• Brian: Mike Woster will need to review any contracts and will then need to sign the contracts with respective vendors

 Action Lars: Send COLO contract to Mike Woster for review
 Action Lars: Send HW and support contracts to Mike Woster for review and finalise as per action above

Vote|Motion:  to vote approve services and HW budget as outlined above with requirement to balance HW support between Intel and AMD cores (test framework WG and members who raised concerns need to sign off – specifically AMD and Oracle) = approved, if revised machine list stays below $58388 and test WG as well as AMD and Intel sign-off

Xen Project Membership

Brainstorm of how we can attract more OEM’s and distros

This is based on an action item I had at the July meeting. Right now, we have Oracle as an OEM and distro on the board. I have been trying to get SUSE for some time to join the board and am not getting anywhere (cost is the issue). Also, I am wondering whether there is a chance to get HP, Dell, … to get additional board members to help address goto-market-issues, which is where we have a weakness compared to KVM (specifically the Open Virtualization Alliance). On the other hand, we do have strengths compared to the OVA in that we have a diverse set of contributors, product companies and large users on the board.

The discussion jumped a little bit (also see the discussion on adding a program manager to the project). The main outcome was that

• Lars: updated on the situation with Suse, who are not likely to join
• Larry: asked whether these are financial or philisophical reasons
• Lars: it looks as if Suse tends to not join projects – e.g. Suse is for example not a member of LEG – but this is not entirely clear. I have been getting mixed messages. On the development side, Suse is increasing their activities with the project though
• The question came up, whether the LF can pro-actively identify and sign up new members via the Program manager role
• Brian: The Linux Foundation is not able to pro-actively identify new members for the project. Even with a Program Manager, this is not really possible because the program manager won’t have sales skills. In addition, opportunities are also more credible coming from members rather than from the LF. Xen Project membership is a business decision: this it is far more powerful if initial membership discussions are made through existing member companies. However, the LF often has the fast path and knows where sensitivities are. For example, the LF can help us to identify whom we bot need to chase. However, the Linux Foundation will follow up on concrete leads from members of the Advisory Board
• Brian: suggested that Advisory Board members could put together a hit list for review after Larry asked whether there is a way for the project to piggy back on activities the LF already does, or whether there is any intelligence that the LF has that will help the project
• Lars: this is what I have done in the past with APM, Rackspace, ARM, GlobalLogic and Qualcomm
• Konrad & Larry: agreed that this was a good idea and that they could follow up to see who their partners are
• Larry: What is the best time to sign up new vendors to projects?
• Brian: Sept / Oct is the best time.
• Thomas: asked whether there are any additional membership requirements to join the Xen Project
• Brian: There is already a requirement that companies are a LF member first: silver minimum

 Action All:: member companies to come up with a list of their partners + contact details by the next board meeting

Brainstorm an extra tier of members for small companies

I recently had a number of discussions with a number of smaller companies which would be willing to join the project if the membership fee for small companies (startups primarily) was lower. Possible candidates for joining would be DornerWorks, Cloudius Systems, Galois, and a few others.

There are some basic concepts that ought to be considered: Breakpoints: for example for Linux Foundation silver membership, membership for companies from 1-99 employees costs $5K instead of $20K for large companies Dilution of voting rights: if we did introduce a membership level which costs less for small companies aligned with the breakpoint system, we ought to ensure that their voting rights are weighted according to the contribution. The standard model within the Linux Foundation would be to assign a number of board seats to members within that I wanted to use the face-2-face meeting to discuss whether there is appetite to look into this further to attract more small companies, and what your specific views on what the structure may be are.

Discussion

• Brian walked through the various options and examples of what other projects do. In other words, there is typically a top tier (which guarantees a board seat) and is called governing membership and one or several lower tiers (where no membership is guaranteed)
• Brian walked through the various breakpoints that exist for Linux Foundation silver membership (e.g. 1-99 employees, etc.)
• Sherry asked whether this could be done by revenue
• Brian: no, because many companies don’t want to / can’t disclose revenue
• A non-governing tier member may not have a board vote at all, or there could be aggregate votes (via a voted representative)
• Lars: asked how this would be bootstrapped. Say we had only one lower tier, then that company may get a board seat because they can vote in themselves
• Thomas: had some concerns whether there was an impact on managaebility of the project, if it got too large. For example, would we need a program manager when we get to a certain size?
• Sherry asked whether there was a requirement for new members to be contributors
• Lars: No
• The underlying issue raised by Sherry was related to companies who do not have an interest in the long term health of the project (or having opposing intentions) joining the board
• Brian: Brian referred to the bad actor clause (in cases where a company joins to disrupt). In theory, in such cases there is the possibility to first warn a bad actor, and if issues persist to remove from the board

Conclusion: as far as I can see, there are no objections to targeting small companies via an extra level of membership. Lars to work with Mike Woster and/or Brian Warner on a concrete proposal for a subsequent board meeting

Budget and early 2015 planning

Budget Sign off Authority

We had a lot of small items that needed to be signed off by the board and I heard from some of you that the board should focus on more high value activities. I discussed with Brian Warner from the Linux Foundation on how other Collaborative Projects solve this. Most of them use different breakpoints for signing off expenses in the budget:

For example – not that the breakpoints ($1.5K and $10K) are made up and we should look at these for expenses less than $1.5K that fit within an approved budget category, the chairman or program manager can sign the expense off for expenses between $1,5K and $10K, the chairman or program manager and 3 board members must sign off for expenses greater than $10K, a majority of the board must sign off I attached the July Budget update for reference to inform that discussion. I did also get a spending report for the first time this year and will try and put these together for the meeting and send out an updated budget report before the meeting, accounting for the recent budget decisions and covering actual income and spending.

Discussion:

• Lars raised the point that recently we have spent a lot of time with budget related items in board meetings and consequently did not have sufficient time to focus on strategic topics
• Larry: we should definitely look into easing this
• Konrad asks how other projects do this
• Brian: The Linux Foundation default is that everything needs to be approved by simple majority. Normally a project would queue up budget related items: in most cases the items would be presented and lazy consensus is applied (aka this are the budget items, does anyone have any objections? If not, there is approval). In some cases the board wants to move quickly and defers decisions to someone it trust to make good decisions.
• Konrad asks whether the process otherwise stays the same
• Brian: the practical side/executive side of process would be the same. Example: buying a $1.5K server – a $1.5 PO is sent to AP – the only difference is that the LF would verify with the chairman instead of checking board meeting minutes
• Brian: if the board was to hire a contractor, the LF would need to also verify that there was someone who gives the contractor direction. Aka there would be an additional check required. If this holds, the LF would recommend board approval.
• Konrad asked whether expenses such as T-shirts fit into the under $1.5K category
• Lars: normally yes

Board members felt, that we should vote for the proposal as laid out in the example above. In other words for expenses less than $1.5K that fit within an approved budget category, the chairman or program manager can sign the expense off for expenses between $1,5K and $10K, the chairman or program manager and 3 board members must sign off for expenses greater than $10K, a majority of the board must sign off

Vote|Motion:  This was approved

Continuing to fund Linux Foundation AR & PR services on behalf of the Project

I had a discussion with Brian and Mike Woster last week and in a nutshell, we need to decide in Q3 whether we want to continue to make use of the AR & PR service from the Linux Foundation. I do believe we should continue do this: from my perspective, the increased visibility for the project is driving growth everywhere else. We are starting to make good progress also on messaging and other areas. What is missing at this stage is a formal working group to help increase the impact of the AR & PR activities.

I wanted to use the face-2-face meeting to get a sense whether we do want continue using this service in 2015.

Discussion:

• Everyone felt that the AR & PR service is good value for the project and that we should continue to use this service in 2015

Looking at other services Linux Foundation services on behalf of the Project

As I stated in the H1 update, I am spending an increasing amount of time on administrative tasks for the project. Today, I am chairman of the Advisory Board as well as de-facto program manager for the project : activities related to the latter are consuming 25-50% of my bandwidth and some of Russell Pavlicek’s bandwidth at the moment. I talked this through with Brian and also discussed with Jeffrey Osier-Mixon who performs the same role for the Yocto Project. In a nutshell, collaborative projects have two ways of doing program management: through an external member resource as we do now through a member of Linux Foundation staff part-time assigned as program manager to the project From my experience in the last year it appears that (1) is significantly less efficient than (2), as I cannot directly access Linux Foundation systems. This creates lots of extra overheads for me and means that I don’t have bandwidth to focus on higher value activities for the project. Albeit, I am not looking for a decision at this meeting, I did want us to start considering whether funding a part-time program manager is something you feel worth considering further down the line. Obviously there will be a budget impact.

Discussion: also see "Brainstorm of how we can attract more OEM’s and distros”

• Konrad asked what the issues were that Lars is seeing
• Lars: some of it is due to Linux Foundation processes not being clearly documented. As a result I often have to ask around and I often get different messages depending on who I ask, Also we had one case, where bills were not paid (e.g. for the Hackathon) and where Lars had to chase AP and where no financial reports were sent for months – again Lars had to chase. Similarly there was a lot of confusion related to booking travel. Some of the rules related to other Linux Foundation services – e.g creative and events services have been changed – without clear communication. Also, the fact that invoices were not issued on time has left us in a situation where we didn’t exactly know how much cash the project had. In a nutshell, most of the issues I have observed come down to undocumented procedures, unclarity about interfaces, and lack of access to Linux Foundation systems
• Lars was also concerned that if he was run over by a bus, the impact on the project would be rather bad at this stage
• Brian recognises the issues and apologised
• Sherry: it appears that some of the terms and conditions had been changed when invoices were re-issued
• Brian: the reason for this was to get everyone onto the same invoicing cycle (different members were invoiced at different times and for different time periods which created confusion). The aim was to reduce complexity and make the process simpler
• Konrad asked what the 15% service charge is for
• Brian: for financial services, invoicing, doing all the paperwork to maintain non-profit status, … everything else is out of scope for the 15% service charge
• Konrad asked what the project would get from a program manager
• Brian: a program manager would perform activities such as
  • Collecting budget reports
  • Producing reports
  • Handle onboarding for new members such as update the website
  • Sponsor recruitment for collaborative projects
  • Anything from speaking to the press … to janitorial work. The main goal of the role is that the project succeeds
  • There is some flexibility, but the role and expectations would need to be established upfront such that the right person is hired. The challenge would be to find a domain expert, that could help the project
• Larry: would chasing possible membership companies be in scope?
• Brian: no, as this is a sales role which is out of scope. Also leads would be significantly m
• Konrad: given that we had a lot of issues with the LF, is there a possibility to get a retroactive discount – e.g. half time program manager at 25% of the cost
• Konrad: voiced dissatisfaction with the level of service provided by the LF
• Brian would need to check with Mike Woster on both items. Brian states that he believes a 25% program manager would not be effective. There is only so much splitting of attention one person can do while staying effective

 Action Brian: Discuss issues raised with Mike Woster
 Action Lars: Sit together with Mike Woster and walk through issues that have been observed Template:Done

Once this is done, for Brian to report back and to work with Lars on making concrete proposals for a subsequent board meeting