My comment should be read as a reflection on the confusion
surrounding all this than as on the OP!
On Mon, 21 Nov 2005, Mathias de Riese wrote:
...
> your confusion, Eygene, arises because there are in fact two uses
> of the word 'BDII' at the moment: The software 'lcg-bdii' and the
> functionality 'bdii'. It gets really bad, because there are two
> functionality which are called 'bdii': The 'site-bdii' and the
> 'top-level-bdii'. The information flow is as follows:
>
>
> -------- ------------- ------------------
> | gris | --> | site-bdii | --> | top-level-bdii |
> -------- ------------- ------------------
> One for One for each One (or more) central
> each re- site. Collec- ones. Usually one for
> source: ting the info each RB. Collects all
> on each about the re- the info about all
> CE,SE,RB, sources of the sites (e.g. lcg-bdii.cern.ch)
> myProxy,.. site.
>
> many fewer one (or few)
>
> The 'site-bdii' used to be called the 'giis'. Since
> the Globus MDS software was originally used for the 'giis'
> functionality and was replaced by the 'BDII' software, it is
> called 'bdii' by many people -- and also by the site-info.def
> variables:
Careful here - I've always heard people use "local-BDII" to refer to the
site GIIS, and "site-BDII" to refer to a "top-level" BDII instance at a
site. Note that the latter (BDII_HOST in site-info.def) is what all the
WNs/UIs use when querying the info system, so there's actually a good
reason for having plenty of them scattered about, rather than one/few.
IMHO "top-level" is a misnomer as there's only one level of BDII; below
that are the site GIISes (albeit implemented with BDII code!).
Thanks
Henry
--
Dr. Henry Nebrensky [log in to unmask]
http://people.brunel.ac.uk/~eesrjjn
"The opossum is a very sophisticated animal.
It doesn't even get up until 5 or 6 p.m."
|