In article <1993Mar25.173738.1848 at gserv1.dl.ac.uk>, ROUCHDA at VAX1.COMPUTER-CENTRE.BIRMINGHAM.AC.UK (Duncan Rouch) writes:
|> What Makes a Successful Biocomputing Service.
|> 1 Introduction: What Success Means in a Biocomputing Service.
|>|> Biocomputing services collectively support four classes of computer-aided
|> methods for biologists; (i) functional- and (ii) evolutionary-sequence
|> analysis, (iii) molecular modelling and (iv) analysis of related information
|> databases, such as literature.
I would argue that this definition is true for many sites, whilst the individual
installation might require different scope. With respect to the licensing
problem, many sites are unable to support molecular modelling, and more
expensive databases like genetic patent data. Similar considerations apply
to genome sequencing support.
|> To give a more practical definition, success in a biocomputing service
|> means helping biologists make the most of the available resources.
|> If adequate resources are not present then an elaboration of what
|> underlies a successful service is needed to argue for these resources.
|> We look at the implications of this definition for the future of academic
|> services here and in a followup discussion .
Success also means a clear distinguishing on what can be supported as a service
and what needs to be individually purchased. At the time where the service
provider cannot afford to support all, it does not mean that a successful service
is not possible or not available. It just means that less features of the
desirable (... by whom? Who decides on which grounds? ) set are going to
|> So, supporting software is an important role of the biocomputing support
|> facility. However, the corollary of the notion of supported software, which
|> includes operating systems, is that some programs are not supported, and
|> clients are made aware of the situation. This maximizes the support that
|> can be given to all clients by a facility with limited resources, since
|> the computer service doesn't have to attempt the impossible by trying to
|> support everything under the sun. The strategy beomes more efficient
|> if biologists subsequently respond by reducing the heterogeneity amongst
|> the software and hardware they use in their labs.
I would add the continuity aspect. Many researchers will insist on feature sets
to be of stable and unchanged nature. E.g., despite the fact that PSQ is now obsolete
and has been replaced by ATLAS, many researchers still work with PSQ and insist
on the feature set to be supported further on. A service provider, following
above guidelines, could think that insufficient learning, and consequently an
intensive user training, can overcome this problem. However, I have found that
'less' is occasionally 'more'. In particular, niche products which resemble very
dedicated features are very expensive in support (both on the user and the
provider side). This is not a plea for mainstream computing but it should be
emphasized that changes are not well appreciated by the user community even if
existing services are being optimized this way.
|> 2.4 Obtaining and Acting on Feedback From Customers.
We have designed the Simplified User Poll and Experience Report Language
(SUPER) which can perform automatic polls in startup procedures. SUPER in
its version 2.0 is available from us and the EMBL fileserver. The publication
of SUPER is submitted.
|>|> 3.2 Implications for a Global Computing Strategy.
|>|> National or regional facilities acrue an advantage through being able to
|> specialize in biocomputing, due to the large number of potential
|> biologist users. In contrast, site-wide facilities generally support
|> different types of computing areas, to accomodate the different types
|> of users at a local site.
|>A recent survey which we conducted in Switzerland (with SUPER, see above) showed
that the local support is very restricted by the budget. Working groups in the
size to 5-15 staff members are not in the position to acquire, and maintain,
knowledge of biocomputing in a broad sense. Therefore, many sites rely
on technical support (i.e., networking, disk and CPU maintenance) to be
supplied locally, but in the end trust in us (the national node) with respect
to general education. The 'real' biocomputing sites who write their
own software and do own research are independent anyway. The conclusions
and findings of the survey are similar to this news posting; the results
will be published.
| Dr. Reinhard Doelz | RFC doelz at urz.unibas.ch |
| Biocomputing | DECNET 20579::48130::doelz |
|Biozentrum der Universitaet | X25 022846211142036::doelz |
| Klingelbergstrasse 70 | FAX x41 61 261- 6760 or 267- 2078
| CH 4056 Basel | TEL x41 61 267- 2076 or 2247 |
+------------- bioftp.unibas.ch is the SWISS EMBnet node ----------------+