Desktop/Server
p@
For the time being, in order to produce a prototype, I think it is
important that we stick with the previous decision to produce a
desktop version of the MeAggregator.
However, I appreciate concerns that it may not be as versatile as
users will want, as in order to use it they will be "tied" to a single
machine. As a step to ameliorate this effect, I believe we can (for
the short term of the 1st prototype) save the FFS model on user's N:
drives, although this may require an extra step for the user to
connect to the N drive if they wish to use MeAggregator on a machine
which is not on the university network.
As I see it (and please correct me if I am misinterpreting the server-
based argument), the difference in approaches revolves around:
Portability between machines - the perception is that users will want
to be able to access their MeAggregator from any machine, but without
having their personal machine left on, and therefore they will need to
have the data and the program available anywhere - thus a server based
implementation is preferable.
Reliability - the perception is that user's own machines are more
prone to outages, thus a server based implementation is preferable.
Ownership - the perception is that users will want to be fully in
control of their data, and thus will not want to have it on an
institutional or third-party server, thus a desktop solution is
preferable.
Synergy and trust - the perception is that users will have more trust
in the application only allowing access to other users that they have
granted permission to if it is on their machine, and that the trust
allows them to make use of the synergy between users.
There are issues relating to caching data, in either scenario. There
is a reasonable chance that users will want to cache all the resources
they tag on the system. This is impractical in a server scenario due
to copyright issues, amongst others, but it is also impractical for
full portability for a desktop based solution. Removing caching
leaves the problem of material which is linked changing.
A possible hybrid might allow caching on the user's personal machine,
and the maintenance of the rest of the model through a server based
option (either providing a central MeAggregator server which runs
individual MeAggregator instances for individuals, or which merely
serves as a repository for the models, and they are transported
between 'desktops')
A server based option has to have a revenue model, or to have
agreement to underwrite the costs of the server by the institution
(unless I am missing another option?) - or it has to be a distributed
system spread across many desktops to provide adequate robustness.
Thoughts?