Paul Gilna writes:
>click! click! hssssss... Whumpfh!
>>couldn't resist it, sorry!
>>>I'm probably crossing all sorts of boundaries here, but I think this
>needs to be said. Regardless of Medline/Genbank/Entrez as the data
>source, the sentiment that I am picking up is that centralised access
>to the data is essential.
>>my belief is that the experience we have here dictates that we must
>dissagree. (much deleted)
Crinkle, crinkle - I believe I've just been flame throwered :-)
Hmm, I find myself caught with my opinion hanging out. I am certainly a
big advocate of distributed database systems and thus do feel a bit chagrined
at being caught advocating a centralyzed system. At the same time I realize
that I am a big fan of some of the centralized services that I use.
Some examples? I really like the world that Wais and Gopher has opened up
to me - and think that such distributed, multi-platform systems are the
way to go.
However some of the centralized systems - such as similarity
searching (Fasta, Blast, Blaze) of sequence databases - is really great and
something I would like to see continue. These centralized services make
a lot of sense to me - because the vast majority of the biological community
only use these services occassionally and as such can't justify the
disk-space/cpu-time/administrative hassle that goes along with maintaining
there own individual copy of genbank/EMBL/Prosite/Swiss-Prot/....
I also understand what Paul is saying about the ever increasing load on
such servers. Perhaps we can learn a lesson from Archie here - it was
a one site resource that started to suffocate under its user load and is
now distributed over 5-10 different nodes (I've lost count). Archie is still
a centralized resource - but has been distributed geographically to better
keep up with user demand. Perhaps the same thing needs to happen with
the sequence databases.
The different nodes of Genbank will have to be funded
but I would think that it is much more effecient than everybody trying to keep
their own copy.
Now perhaps there should be different nodes of NLM for Greatful Med access?
I know I know - many libraries run there own little version of Medline, but
given the choice - I'd rather use Greatful Med - the interface is much better
than many of the medline systems that I've seen and the database has a greater
coverage of the literature. This may not even be an issue in a few years
as I suspect that private companies will step in to fill this literature
searching void. I know of at least one company that is creating a database
from about 12,000 journals which one will be able to search, and download
the desired articles (not just the reference). (Please don't ask me for
further details on this project you've just read almost everything I know)
I know there are lots of cans of worms being opened here - but my point is
that there are places for both the widely distributed systems (like wais and
gopher) and for more centralized resources - which may be spread across a few
nodes for load considerations.
danj at welchgate.welch.jhu.edu
Disclaimer: These are my opinions - you can have them if you want them but
they may not be equal to those of my employer.