[lopsa-discuss] Interruptions coverage...

[Betsy Schwartz]

At Genuity, I was part of a team that had 6-12 people over time. We had an 
on-call rotation where one person was primary and one person was backup for 
a full week. The oncall person did little but respond to tpages , around 
the clock. The backup oncall person handled the ticket queue, which was 
often quite large. They did the quick tickets themselves and handed off 
others to folks who had particular areas of responsibility. Plus, they 
sometimes got paged by the primary oncall or engaged if the primary didn't 
respond within 15 minutes.

The down side of this was that everyone had long-term  projects to work on, 
and both the primary and backup oncall ended up sandbagged for an entire 
week at a time. This could play hell with deadlines. The upside was that it 
was always pretty clear whose responsibility the queue was. There was often 
a bit of friction about how many open tickets were left in the queue when 
it was handed over to the next person; I'd wonder if this could be an issue 
with very frequent turnover of responsibility.

A good ticket system is a MUST. We had a company-wide Vantive system at 
Genuity; when I was at UMB we installed RT (or its' predecessor). (Harvard 
has Remedy, which I use to interact with the central IT guys but not in our 
group)

At my current job I'm the only unix admin, although I have some backup from 
the two web folks and a part-part-time assistant. We have four helpdesk 
guys and a  PC-based ticket system with a PC-only interface  (HEAT , 
www.targetfour.com ). I find it sorely lacking,  but I'm the only one on 
the team who has experience with more full-featured ticket systems. ( I 
abandoned my attempt to interest folks in RT after we hired a new helpdesk 
manager with extensive experience with HEAT. )

Things I can do as a solo person: we have a web-based interface to our user 
management system, so we're working on  a combination of better 
documentation and adding tools to the web interface so that the helpdesk 
folks can handle more calls.
I also work from home one day a week, and I do schedule uninterrupted time 
for myself and close my door. We use IM but the interruptions aren't that 
frequent.

I'm still reading Tom's book but I read Getting Things Done and really like 
that system, especially the idea of spending time weekly to break projects 
down into "next actions". Of course, being a sysadmin means getting 
bazillions of "Next Actions". I sorted mine out into office, system, 
documentation, and "short" (things that don't involve root, mostly) and I 
blocked out time during the week to work on each. Of course I find excuses 
not to do documentation.....





Betsy Schwartz                                                    email: 
betsys at gsd.harvard.edu
Unix Systems Administrator,CRG                           voice: 617-495-5947
Harvard Graduate School of Design                         fax:    617-496-5866