[AG-DEV] Node-listing data, course of action
Brian Corrie
bcorrie at sfu.ca
Thu May 31 17:03:28 CDT 2007
Hello all,
Time for me to add my $0.02 cents worth.
Let me start with raising a point. In my view we are talking about two
different things here, maintaining an international list of AG nodes and
a QA process. These two are independent in my mind and having one does
not necessarily depend on the other. It would be a good thing to have
them tightly linked (cross-referenced in some way) but they don't need
to and probably shouldn't be a single entity...
Jason Bell wrote:
> G'day all
>
> Progressing with the sharing of node-listing data, particularly in
> regards to the QA process, may I suggest the following comments:
>
> * Currently it appears that only the UK people have the need for
> their own DB/system;
I don't think that is the case... We only started a discussion of a
unified schema for node listings a little over a week ago, or a bit
longer for those who were at the AG retreat. Heck, I have email in my
InBox that I consider important that is several months old!!! (but no,
we won't digress into a discussion of my work habits 8-)
We are looking at this in some detail, and I can see we will likely need
our own database (in fact we already have one that we are testing). Our
anticipated use of this data base will require more information than
what a "simple" node listing would typically have, and that information
will be very specific to our needs. That is, it doesn't make sense for
the global node listing to reflect that information.
I suspect this will be true for many groups/consortias/countries...
There are three important things to consider here:
1) What is the definitive data that is important for a global node listing.
2) Where is that information stored
3) How is that information utilized (e.g. displayed at a web site).
We have a good first pass at what 1) might be from what you sent from
the accessgrid.org node listing. With refinement I think we can probably
come to some agreement on what this information should be.
In many ways, 3) is easy. Once we have the information in a database
somewhere, the information can be displayed in a variety of ways by a
variety of groups.
I think 2) is harder. The crux of the problem is where is the definitive
data for a given node going to be, who is going to keep that data up to
date, and how are others going to be able to make use of that data.
The needs from the perspective of 2) is different for different types of
organizations. If you are a university and have a couple of nodes, then
sure, put your nodes up on a central site. You certainly don't want to
build a database yourself. So we clearly need that capability. But what
if you are a consortium that is providing a service for a large number
of nodes (AGSC, WestGrid). Does it make sense for WestGrid to enter all
of our nodes into the accessgrid.org node listing site. Especially when
we already have a database of nodes?
It seems to make more sense to me to pull information from the regional
database and push that to the accessgrid.org site. The regional database
is going to be kept up to date and is the definitive description of the
node. Although I am not an expert, I would think that it would be
relatively easy to create a web service or some such beast that
transforms our description of a node to the global description for
either updating the central data base or possibly just rendering to a
web page.
The other option is of course to go the other way, with the definitive
data at accessgrid.org and regional organizations pull information from
there. That doesn't make a lot of sense to me. We will still need our
own database, so why are we storing important information to us at a
remote location...
If we aren't pushing data between these databases, then we will end up
in the same position we are in now. We will run our own stuff for node
listings, bookings, contact people, and when we get reminded by Jason we
will up date our node information at the accessgrid.org site. 8-) I know
which node listing will remain the most up to date from our sites...
> * With this in mind, I will proceed with QA'ing nodes and regional
> testers so that the AG's quality can be improved;
As I said, I think we should be careful here. I feel that the QA process
and node listings are separate issues... Lets try to keep what we are
talking about clear...
Brian
More information about the ag-dev
mailing list