Re: site servers

Bob Carlitz (bob@hamlet.phyast.pitt.edu)
Fri, 28 Mar 1997 01:01:57 -0500 (EST)


> Let me respond to one piece of this question --
>
> > 	* What pitfalls are there to server implementation 
> > 	  and maintenance?  What works and what doesn't work
> > 	  in your experience?
>
> In our work in the DoDEA schools in Hanau, there was been an email capacity
> (cc:mail) for over a year. Just in the past few months, a LAN has been
> finished with 8 drops per classroom and a minimum of 2 computers per
> classroom. All teachers now have Internet (incl email) access to their
> desktops, and students have access in their classrooms as well as labs and
> the library.

	How well does this scale?  Is cc:mail priced in terms of
	the number of users, or does it come with a flat license?
	If a school has to pay additional money for additional
	e-mail use, they probably will restrict student use in
	the long term.  I know that some commercial e-mail packages
	have this problem, but I'm not sure about cc:mail.
 
> The administrative load has increased enormously over this period of time,
> and the district person assigned to meet this need has had to work past any
> normal limits to meet the demands. A sys admin person has been hired by DoDEA
> and will be starting early in April; we all have our fingers crossed that she
> will have the requisite skill set.  

	This leads to another question about scaling.  In Common
	Knowledge: Pittsburgh we have deployed servers at all of
	our school sites.  But the architecture of these servers
	is such that most system maintenance can be done centrally.
	This allows a single skilled system administrator to
	handle dozens of school-based servers.  Each school needs
	to provide someone to handle user administration chores -
	account creation, backups and quotas - but these tasks
	require far less training and are less intrinsically
	stressful than strict system administration.

	This architecture works well in the setting of an urban
	school district.  But since the remote maintenance is
	carried out over the network, the central system
	administrator could be located in another city, and
	the extent to which administrative chores were
	aggregated could also go beyond the boundaries of a
	single municipality.  In the DoDEA context this could
	allow schools around the world to share a single
	system administrator.  This is an unusual-sounding idea,
	but given stable network links, it's not completely
	unreasonable.

	Bob Carlitz