Cloudscaling’s Randy Bias at OpenStack SV: “The Lie of the Benevolent Dictator”

Cross-posted from the Mirantis blog

By Nick Chase

Everybody’s got ideas as to how to make OpenStack better. And that’s the problem, according to Cloudscaling’s Randy Bias.  Bias took to the stage at yesterday’s OpenStack Silicon Valley with a call to arms, explaining what he thought was wrong with the way OpenStack is being run, and why he thinks the return of the Central Planning Committee might be the answer.

There’s been a lot of talk about why OpenStack needs a benevolent dictator, ala Linux, but Bias immediately squashed that kind of talk.  To him, the problem isn’t leadership.  Openstack has leadership, he pointed out, or at least a governance model.

No, the problem with OpenStack is that nobody, not the Board of Directors, which sees itself as providing general guidance and oversight, nor the Technical Committee, which sees itself as responsible for pushing out the twice yearly updates, no, nobody is responsible for making sure that OpenStack has a product strategy, or at least a vision of where it should go.

“‘Being OpenStack’ is not a strategy,” he says.  He feels that it doesn’t mean anything, and whatever it does mean gets watered down further every time a new project gets added.  While Defcore and Refstack help, they don’t solve the problem, in his view.

The problem, in his view, is that like anything you’re trying to put in a customer’s hands, OpenStack is a product, and a strategy won’t “’emerge’ from a group”. He likened it to the story of the blind men and the elephant. One felt the leg and thought it was a tree, one felt the trunk and thought it was a thick rope, and so on.  Only in Bias’ view, the elephant is getting bigger, impeding OpenStack’s progress and potentially threatening its success.

Bias believes OpenStack needs the following:

  • To Manage OpenStack like a product

  • Long term vision and planning

  • Corporate independance

  • To focus on end user needs and requirements

  • The Board and TC working more closely

  • Architectural oversight and leadership

Most or all of those items seem reasonable, but how does he propose the community achieve it?  Taking a cue from Amazon Web Services, which has a small architectural review board and a team that does product management, he suggests two new structures:

 

Architecture Review Board

  • 2-4 year terms (so they can see the long view)

  • Small number of seats

  • Wide set of domain experience

  • Remit:

    •  Work with the TC on Blueprints

    •  Help the Product Managers validate priorities

Product Management

  • Not elected

  • 1+ per program

  • Specific domain experience

  • Wok for the Foundation

  • Annual 360 degree feedback

  • Remit:

    •  Priority Setting

    •  Work with PTL

However OpenStack moves forward, Bias says, any successful path requires the creation of ownership of vision and strategy in somebody. To avoid insurmountable challenges, “We have a collective responsibility to fix this.”

Leave a Reply

Your email address will not be published. Required fields are marked *