weblogs.com Project Follow-up

Dave Winer had to bring down service on his free weblogs.com accounts recently. It turned into a bumpy ride for a while, and perhaps could have been handled better. Dave says,

One of the things that came from this is a process for shutting down a free service. Let’s draft a common document that explains to people how to do it, and really consider the problems.

We aren’t breaking any new ground here. Systems changes, quiescing a production environment, communication with stakeholders, migration of data across servers… all these matters and more are simply part of a systems project management discipline that many of us around here actually earn a living practicing. I’ve been seriously thinking of using what I’ve observed about this project’s initiation as a case study.

Dave has brought forward a serious proposition. Regarding his recent success at bringing down the free weblogs.com service without leaving his users high and dry, he says:

I don’t want an argument, I want to learn, based on two things, the benefit of time – it’s almost two weeks since the server went off the air, and the sites are back up and redirected now. And second – the emotions have had time to die down, to the point where the issues are not emotional.

I think the problems Dave faced in early June, his movement toward resolving the problems, the reactions that formed in what we might call an adjacent user community, the support that coalesced, the actions that were taken, and the final resolution of matters for Dave and the weblogs.com users are worthy of a case study in systems administration and project management. We all might have a chance to learn something.

The discipline surrounding project management scales. Whether you're a one man band or you're leading an IT shop of three or four hundred employees, whether you want to bring up a web server serving a few bloggers at your house on the end of a DSL circuit or establish a colo service for fortune 500 companies, there are steps you can follow to improve your chances of success in terms of time, cost, and ultimately customer satisfaction.

I don't think scale is terribly important here. There were only 3000 sites and they were being hosted for free so there's no revenue stream complicating our assessment, no ROI, no hurdle rate, none of the business matters that can muddy the water. Staffing was an issue. If Dave couldn't do it, it wasn't going to get done. Dave's communications choices and his perspective on who the stakeholders were seemed to be a disruptive factor in the beginning. I think his "guest" or "gift" accounts, the users who had received a free service, were his primary stakeholders and he was accurate in his assessment that they might suffer a little from a disruption of service, but by July 1 he was committed to give anyone who requested it a copy of his or her weblog.

Now it could be argued that the terms of service for the free accounts were such that just pulling the plug and consigning the blogs to the bit bucket was Dave's right. For me, that's not the point. That discussion follows later, after the creation of a problem statement, or a project initiation. That discussion is part of an evaluation of alternatives. That was one alternative to assess among others as part of a disciplined systems project management process. And a successful project includes a thoroughgoing stakeholder analysis... who were the others who might be affected by the disruption, how might they wish to be informed, what issues did they face? If public opinion is a factor, how do we deal with that?

I think Doc's IT Garage is the perfect forum for mulling over the systems project issues, the choices, the level of planning that we might want to do if faced with this kind of problem in the future. Does anybody else want to discuss the systems management issues Dave was facing a few weeks ago when he brought the weblogs.com service down?

(Incidentally, this forum permits moderation and I'll try to provide that as a service if people want to discuss things here.)