There were two relevant case studies presented, but before that, the keynote was by Dylan Larson from Intel on virtualisation.
At one point I nearly reached MEGO (my eyes glazed over) because I'm not in the least bit familiar with chip technology and it got a teensy bit too technical. But on the whole, I managed to follow much of the presentation. The issues targeted by virtualisation were server sprawl, power & cooling requirements (where this often costs 1/2 as much as the hardware needing to be powered and cooled), ooperating costs and space crunch (or lack of space).
Anyway, back to the case studies... key messages
- problems with lack of local knowledge and experience
- government sites were very helpful for guidelines on record management
- single trainer used for entire company - same person who works on service desk - this was cited as being a good thing
- Posed the question that it might have been better (although more painful initially) to make a big change for staff, rather than try to do it incrementally
- There was a false confidence in user acceptance - difficulties didn't come to light until network drives were made read-only
- "the earlier technical staff brought in the better" - this did not happen until much later
- Good to keep the focus on the business needs
- Keep dialogue with vendor open
- communicate every step
- secure support past the end of the project
- require developers to work on site with staff to develop knowledge in house
- do not customise too much - take it out of the box if you can
- users are not motivated or understand why metadata is important - (consider metadata checking be done centralyy)